interaction_id stringlengths 9 9 | session_id stringlengths 6 6 | user_id stringclasses 438
values | timestamp_utc stringdate 2025-02-01 02:22:18 2025-04-30 22:23:19 | split stringclasses 3
values | channel stringclasses 5
values | use_case stringclasses 6
values | country_code stringclasses 10
values | region stringclasses 3
values | account_tier stringclasses 3
values | segment stringclasses 3
values | instruction_template stringclasses 6
values | model_provider stringclasses 6
values | model_name stringclasses 8
values | temperature float64 0 1.2 | top_p float64 0.71 1 | max_tokens int64 256 2.05k | prompt_tokens int64 13 1.91k | completion_tokens int64 5 1.12k | total_tokens int64 29 2.75k | latency_ms int64 156 5.62k | cost_usd float64 0 1.05 | response_quality_score float64 0.09 1 | is_failure bool 2
classes | failure_type stringclasses 7
values | hallucination_flag bool 2
classes | toxicity_flag bool 2
classes | safety_block_flag bool 2
classes | formatting_error_flag bool 2
classes | tool_error_flag bool 2
classes | latency_timeout_flag bool 2
classes | tool_calls_count int64 0 3 | tools_used stringclasses 65
values | user_feedback_score float64 -2 2 ⌀ | user_feedback_label stringclasses 4
values | user_reported_issue bool 2
classes | retry_index int64 0 3 | repair_strategy stringclasses 7
values | final_resolution_state stringclasses 3
values | business_impact_tag stringclasses 4
values | request_text stringlengths 63 301 | response_text_snippet stringlengths 180 263 | tokens_per_second float64 27.4 7.66k | date_utc stringdate 2025-02-01 00:00:00 2025-04-30 00:00:00 | hour_of_day_utc int64 0 23 | day_of_week int64 0 6 | is_weekend bool 2
classes | is_peak_hour bool 2
classes | prompt_to_completion_ratio float64 0.02 208 | request_text_template stringclasses 140
values | instruction_text stringclasses 6
values | user_feedback_observed bool 2
classes | prompt_id stringclasses 36
values | response_text_snippet_len_chars int64 180 263 | response_text_snippet_is_truncated bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
INT008701 | S00140 | U00081 | 2025-02-12T11:15:27Z | test | web_app | data_analysis | FR | EMEA | pro | team | code_assistant_secure | meta | llama-3.1-70b | 0.109 | 0.807 | 512 | 1,052 | 347 | 1,399 | 649 | 0.2798 | 1 | false | none | false | false | false | false | false | false | 2 | email_send|workflow_orchestrator | null | no_feedback | false | 0 | none | auto_resolved | medium | I'm working on something and need help: Help me compute the conversion rate per marketing channel and highlight the top 3 performers. I'd like a practical, engineering-focused answer. | Let's break this down into actionable steps. Goal: handle cost per request reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Start by... | 2,155.624037 | 2025-02-12 | 11 | 2 | false | true | 3.0317 | Help me compute the conversion rate per marketing channel and highlight the top 3 performers. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 4e275efe35fdc513983e7d3f3ad32b50cc391969 | 237 | true |
INT008702 | S00795 | U00120 | 2025-02-21T10:08:51Z | test | internal_tool | coding_assistant | GB | EMEA | free | individual | analytics_helper | mistral | mistral-large | 0.234 | 0.867 | 1,536 | 573 | 567 | 1,140 | 1,299 | 0.1368 | 0.797 | false | none | false | false | false | false | false | false | 1 | workflow_orchestrator | 2 | positive | false | 0 | none | auto_resolved | high | I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. Please explain step by step. | This is a common pattern; the fix is straightforward. Goal: handle caching strategy reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Start by writing a... | 877.598152 | 2025-02-21 | 10 | 4 | false | true | 1.0106 | I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | 54f94ac4f6df7ccc4808605ae4fb839b7a1fbfb2 | 257 | true |
INT008703 | S01595 | U00040 | 2025-02-17T08:49:15Z | train | mobile_app | customer_support | IN | APAC | pro | individual | internal_policy_qa | google | gemini-1.5-pro | 0.481 | 0.902 | 768 | 1,520 | 677 | 2,197 | 3,284 | 0.70304 | 0.704 | false | none | false | false | false | false | false | false | 1 | sql_query | 1 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Please explain step by step. | Below is a practical plan you can execute. Goal: handle slow responses reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use schema validation... | 669.001218 | 2025-02-17 | 8 | 0 | false | false | 2.2452 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 03bf57d7adc83026d525c8192379aa203b71e1eb | 246 | true |
INT008704 | S01326 | U00389 | 2025-04-01T19:21:30Z | train | web_app | coding_assistant | EG | EMEA | free | individual | general_assistant | openai | gpt-4.1 | 0.28 | 0.848 | 2,048 | 845 | 625 | 1,470 | 985 | 0.588 | 0.54 | true | safety_block | false | false | true | false | false | false | 0 | none | -2 | negative | false | 3 | manual_post_edit | escalated_to_human | medium | Quick question: Follow-up: the previous reply did not fully solve the issue. Write a small example that shows how to handle exceptions properly in JavaScript. A short, clear answer would be ideal. | Let's turn this into a reliable workflow. Let's stabilize it. Signal: safety_block while working on FastAPI endpoint design. Quick diagnosis: Route to a safe-help template and suggest benign resources or best practices. Recommended... | 1,492.385787 | 2025-04-01 | 19 | 1 | false | false | 1.352 | Follow-up: the previous reply did not fully solve the issue. Write a small example that shows how to handle exceptions properly in JavaScript. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | 3ddc6acb5986d68a027a4c760a5adfbe894b8deb | 234 | true |
INT008705 | S00089 | U00070 | 2025-03-24T21:24:56Z | test | web_app | customer_support | US | AMER | free | enterprise_team | internal_policy_qa | meta | llama-3.1-70b | 0.848 | 0.847 | 512 | 1,084 | 413 | 1,497 | 1,322 | 0.2994 | 0.571 | true | tool_error | false | false | false | false | true | false | 3 | knowledge_lookup|sql_query|workflow_orchestrator | -1 | negative | true | 0 | switch_model_family | escalated_to_human | medium | Quick question: Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Can you walk me through a good solution? | We can make this robust with a few changes. Let's stabilize it. Signal: tool_error while working on data export request. Quick diagnosis: Separate tool telemetry from model telemetry so you can see which component failed. Recommended... | 1,132.375189 | 2025-03-24 | 21 | 0 | false | false | 2.6247 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 03bf57d7adc83026d525c8192379aa203b71e1eb | 236 | true |
INT008706 | S00738 | U00063 | 2025-03-18T11:02:12Z | train | web_app | customer_support | BR | AMER | free | individual | internal_policy_qa | mistral | mistral-large | 0.161 | 0.873 | 1,024 | 1,053 | 394 | 1,447 | 889 | 0.17364 | 0.747 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Quick question: A user says they were double charged this month. Help me draft a clear explanation and next steps. Can you walk me through a good solution? | Here's the clean way to approach it. Goal: handle login issues reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Add telemetry fields for... | 1,627.671541 | 2025-03-18 | 11 | 1 | false | true | 2.6726 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 241 | true |
INT008707 | S00997 | U00298 | 2025-04-28T17:42:19Z | val | mobile_app | customer_support | FR | EMEA | free | enterprise_team | internal_policy_qa | mistral | mistral-large | 0.781 | 0.921 | 768 | 525 | 446 | 971 | 835 | 0.11652 | 0.804 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | high | I need a short, friendly reply to a customer who is asking about the status of their refund. A short, clear answer would be ideal. | We can make this robust with a few changes. Goal: handle missing feature reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for missing feature and add tests before... | 1,162.874251 | 2025-04-28 | 17 | 0 | false | true | 1.1771 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 230 | true |
INT008708 | S00547 | U00426 | 2025-04-24T10:19:29Z | train | web_app | coding_assistant | FR | EMEA | free | team | support_strict | google | gemini-1.5-pro | 0.185 | 0.91 | 1,536 | 1,028 | 189 | 1,217 | 1,962 | 0.38944 | 0.553 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | null | no_feedback | false | 1 | regenerate_with_lower_temp | escalated_to_human | medium | I'm working on something and need help: Follow-up: the previous reply did not fully solve the issue. Can you refactor this function to be more readable and efficient while keeping the same behaviour? | Here's what I'd implement in production. Let's stabilize it. Signal: tool_error while working on unit test mocking. Quick diagnosis: Implement a fallback path when the tool is degraded. Recommended actions: - Set alert thresholds on... | 620.285423 | 2025-04-24 | 10 | 3 | false | true | 5.4392 | Follow-up: the previous reply did not fully solve the issue. Can you refactor this function to be more readable and efficient while keeping the same behaviour? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | da34aac46ee8d6d44012043be92e33884c6b4920 | 235 | true |
INT008709 | S00167 | U00299 | 2025-04-11T22:44:07Z | test | web_app | internal_qa | US | AMER | free | individual | support_strict | local | custom-local-8b | 0.543 | 0.914 | 1,536 | 679 | 731 | 1,410 | 1,184 | 0.00564 | 0.703 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | high | Quick question: Check this draft announcement for any confusing or inconsistent statements. Can you walk me through a good solution? | Here are the guardrails I'd put in place. Goal: handle provider degradation reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure... | 1,190.878378 | 2025-04-11 | 22 | 4 | false | false | 0.9289 | Check this draft announcement for any confusing or inconsistent statements. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | true | 4e3bb3b7c597daf93de3f611632f4c56cefdec0e | 219 | true |
INT008710 | S01570 | U00001 | 2025-02-24T08:06:55Z | val | web_app | brainstorming | IN | APAC | free | individual | support_strict | meta | llama-3.1-70b | 0.684 | 0.931 | 1,024 | 249 | 811 | 1,060 | 418 | 0.212 | 0.847 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | abandoned | medium | I'm working on something and need help: I need naming ideas for an internal tool that tracks experiments and model runs. | Here are the guardrails I'd put in place. Goal: handle incident response automation reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 2,535.885167 | 2025-02-24 | 8 | 0 | false | false | 0.307 | I need naming ideas for an internal tool that tracks experiments and model runs. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | true | e3da6c15e87a6fab71f2d2b735988233f5d1f30b | 231 | true |
INT008711 | S00229 | U00180 | 2025-03-18T00:57:10Z | train | mobile_app | customer_support | IN | APAC | pro | individual | analytics_helper | meta | llama-3.1-70b | 0.445 | 0.821 | 768 | 26 | 614 | 640 | 1,837 | 0.128 | 0.629 | false | none | false | false | false | false | false | false | 3 | email_send|http_request|knowledge_lookup | null | no_feedback | false | 0 | none | abandoned | critical | Quick question: A user says they were double charged this month. Help me draft a clear explanation and next steps. | Here's the clean way to approach it. Goal: handle login issues reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 348.394121 | 2025-03-18 | 0 | 1 | false | false | 0.0423 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 598eba0ae15195966a9835b68c542f0b34fb88e4 | 210 | true |
INT008712 | S01135 | U00077 | 2025-04-21T18:39:18Z | train | slack | brainstorming | US | AMER | enterprise | individual | support_strict | google | gemini-1.5-pro | 0.737 | 0.922 | 1,536 | 1,627 | 183 | 1,810 | 2,243 | 0.5792 | 0.969 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 1 | regenerate_with_lower_temp | auto_resolved | high | Can you help me debug this? Follow-up: the previous reply did not fully solve the issue. This is coming from an internal team in our company. Give me ten creative ideas for a YouTube series about learning data science from scratch. I'd like a practical, engineering-focused answer. | Let's break this down into actionable steps. Goal: handle LLMOps dashboards reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use schema... | 806.954971 | 2025-04-21 | 18 | 0 | false | true | 8.8907 | Follow-up: the previous reply did not fully solve the issue. This is coming from an internal team in our company. Give me ten creative ideas for a YouTube series about learning data science from scratch. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | false | e3da6c15e87a6fab71f2d2b735988233f5d1f30b | 240 | true |
INT008713 | S00958 | U00405 | 2025-04-03T22:56:23Z | train | internal_tool | content_writing | CA | AMER | free | team | support_strict | local | custom-local-8b | 0.66 | 0.89 | 1,024 | 999 | 767 | 1,766 | 2,348 | 0.00708 | 0.488 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | -2 | negative | true | 2 | switch_model_family | auto_resolved | low | Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. Can you walk me through a good solution? | Here are the guardrails I'd put in place. Signal: tool_error while working on tool calling. Quick diagnosis: Implement a fallback path when the tool is degraded. Recommended actions: - Add a... | 752.129472 | 2025-04-03 | 22 | 3 | false | false | 1.3025 | Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 75b7366387b6cf4f5631ec9dd298a359e5bb9c1d | 193 | true |
INT008714 | S00422 | U00407 | 2025-04-05T05:06:18Z | train | internal_tool | customer_support | US | AMER | free | team | creative_writer | openai | gpt-4.1 | 0.557 | 1 | 1,024 | 33 | 291 | 324 | 770 | 0.1296 | 0.84 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | escalated_to_human | low | I'm stuck on this and need some guidance: I need a short, friendly reply to a customer who is asking about the status of their refund. A short, clear answer would be ideal. | I'll keep this concrete and measurable. Goal: handle slow responses reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Add telemetry fields... | 420.779221 | 2025-04-05 | 5 | 5 | true | false | 0.1134 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 7a490bfafd7436aea43a528e9fe4cb6db10b9a6b | 243 | true |
INT008715 | S00206 | U00273 | 2025-04-28T08:12:41Z | train | api | content_writing | AE | EMEA | free | team | analytics_helper | meta | llama-3.1-70b | 0.808 | 0.929 | 1,024 | 756 | 441 | 1,197 | 651 | 0.2394 | 0.449 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | -2 | negative | false | 0 | add_constraints_to_prompt | escalated_to_human | high | I'm working on something and need help: Turn these bullet points into a short, engaging blog introduction about learning machine learning. I'd like a practical, engineering-focused answer. | We can make this robust with a few changes. Signal: tool_error while working on tool calling. Quick diagnosis: Separate tool telemetry from model telemetry so you can see which component failed.... | 1,838.709677 | 2025-04-28 | 8 | 0 | false | false | 1.7143 | Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 9716fa3b2c6fe39ac0f75ddaad781dfa757a0de9 | 197 | true |
INT008716 | S00569 | U00063 | 2025-02-21T21:45:35Z | train | mobile_app | internal_qa | BR | AMER | free | individual | general_assistant | meta | llama-3.1-70b | 0.232 | 0.893 | 1,536 | 249 | 450 | 699 | 2,807 | 0.1398 | 0.908 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | critical | Context for you: Explain our refund policy in simple language that a new support agent can remember. Can you walk me through a good solution? | Here's the clean way to approach it. Goal: handle token spikes reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. -... | 249.020306 | 2025-02-21 | 21 | 4 | false | false | 0.5533 | Explain our refund policy in simple language that a new support agent can remember. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | true | 7cd4f5f09ad56cc7a4cac3fea1f795d4e2ccc742 | 217 | true |
INT008717 | S00508 | U00374 | 2025-02-02T17:45:32Z | train | internal_tool | customer_support | BR | AMER | pro | enterprise_team | analytics_helper | openai | gpt-4.1 | 0.304 | 0.95 | 1,536 | 365 | 298 | 663 | 1,044 | 0.2652 | 0.526 | true | tool_error | false | false | false | false | true | false | 1 | code_run | 0 | neutral | false | 0 | add_constraints_to_prompt | escalated_to_human | low | Can you help me debug this? A user says they were double charged this month. Help me draft a clear explanation and next steps. Please explain step by step. | Below is a practical plan you can execute. Signal: tool_error while working on data export request. Quick diagnosis: Validate tool arguments before calling the tool; reject invalid calls early.... | 635.057471 | 2025-02-02 | 17 | 6 | true | true | 1.2248 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 598eba0ae15195966a9835b68c542f0b34fb88e4 | 196 | true |
INT008718 | S00315 | U00048 | 2025-03-16T16:02:09Z | train | web_app | customer_support | IN | APAC | free | enterprise_team | internal_policy_qa | meta | llama-3.1-70b | 0.429 | 0.907 | 768 | 535 | 54 | 589 | 751 | 0.1178 | 0.964 | false | none | false | false | false | false | false | false | 3 | email_send|sql_query|vector_search | null | no_feedback | false | 0 | none | auto_resolved | low | I need a short, friendly reply to a customer who is asking about the status of their refund. Can you walk me through a good solution? | Here are the guardrails I'd put in place. Goal: handle data export request reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Add telemetry fields... | 784.287617 | 2025-03-16 | 16 | 6 | true | true | 9.9074 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 249 | true |
INT008719 | S00613 | U00022 | 2025-04-24T03:05:46Z | val | web_app | coding_assistant | EG | EMEA | pro | team | support_strict | openai | gpt-4o-mini | 0.152 | 0.849 | 1,024 | 939 | 629 | 1,568 | 626 | 0.03136 | 0.49 | true | hallucination | true | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | add_constraints_to_prompt | auto_resolved | low | I'm working on something and need help: Can you refactor this function to be more readable and efficient while keeping the same behaviour? | I'll keep this concrete and measurable. Signal: hallucination while working on API retry logic. Quick diagnosis: Add grounding: retrieve evidence, require citations, and verify key facts.... | 2,504.792332 | 2025-04-24 | 3 | 3 | false | false | 1.4928 | Can you refactor this function to be more readable and efficient while keeping the same behaviour? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | da34aac46ee8d6d44012043be92e33884c6b4920 | 190 | true |
INT008720 | S00877 | U00401 | 2025-04-19T04:03:30Z | train | internal_tool | customer_support | BR | AMER | free | team | support_strict | mistral | mistral-large | 0.246 | 0.964 | 1,536 | 69 | 46 | 115 | 893 | 0.0138 | 0.968 | false | none | false | false | false | false | false | false | 3 | email_send|knowledge_lookup|vector_search | null | no_feedback | false | 0 | none | auto_resolved | low | Can you help me debug this? Our customer is getting an error when trying to log in to their account. Can you help me debug this? | This is a common pattern; the fix is straightforward. Goal: handle slow responses reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use... | 128.779395 | 2025-04-19 | 4 | 5 | true | false | 1.5 | Our customer is getting an error when trying to log in to their account. Can you help me debug this? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 244 | true |
INT008721 | S01311 | U00431 | 2025-02-11T15:13:35Z | train | web_app | coding_assistant | US | AMER | free | enterprise_team | code_assistant_secure | local | custom-local-8b | 0.213 | 0.968 | 1,024 | 1,058 | 599 | 1,657 | 2,297 | 0.00664 | 0.935 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | medium | I'm working on something and need help: Write a small example that shows how to handle exceptions properly in JavaScript. I'd like a practical, engineering-focused answer. | Below is a practical plan you can execute. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for async I/O pitfalls and add... | 721.375707 | 2025-02-11 | 15 | 1 | false | true | 1.7663 | Write a small example that shows how to handle exceptions properly in JavaScript. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | 52eadb52b3319d2dab569ce8fb653a0bee56a90f | 222 | true |
INT008722 | S00221 | U00077 | 2025-04-23T16:02:38Z | train | api | customer_support | US | AMER | enterprise | individual | code_assistant_secure | google | gemini-1.5-pro | 0.435 | 0.802 | 1,024 | 1,381 | 907 | 2,288 | 993 | 0.73216 | 0.756 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | medium | A user says they were double charged this month. Help me draft a clear explanation and next steps. | Here's what I'd implement in production. Goal: handle account verification reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure... | 2,304.128902 | 2025-04-23 | 16 | 2 | false | true | 1.5226 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 1a8e13c1d6c8047debb9a230bf3ae08524ccb884 | 218 | true |
INT008723 | S00864 | U00414 | 2025-04-16T18:53:47Z | train | api | brainstorming | GB | EMEA | free | team | support_strict | mistral | mistral-large | 0.279 | 0.896 | 768 | 511 | 518 | 1,029 | 796 | 0.12348 | 0.626 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | medium | Quick question: Brainstorm taglines for a productivity app that helps engineers stay focused. I'd like a practical, engineering-focused answer. | We can make this robust with a few changes. Goal: handle LLMOps dashboards reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Add telemetry fields... | 1,292.713568 | 2025-04-16 | 18 | 2 | false | true | 0.9865 | Brainstorm taglines for a productivity app that helps engineers stay focused. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | true | e3da6c15e87a6fab71f2d2b735988233f5d1f30b | 250 | true |
INT008724 | S01006 | U00151 | 2025-03-13T16:32:12Z | train | slack | data_analysis | CA | AMER | pro | individual | internal_policy_qa | anthropic | claude-3.5-sonnet | 0.586 | 0.882 | 1,536 | 856 | 497 | 1,353 | 1,234 | 0.43296 | 0.797 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Can you suggest a few plots to explore churn risk in this dataset of subscriptions? Can you walk me through a good solution? | We can make this robust with a few changes. Goal: handle token usage reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Start by writing a small... | 1,096.43436 | 2025-03-13 | 16 | 3 | false | true | 1.7223 | Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | 62646e5a7026d4f68ab1867a730451326aa3b664 | 247 | true |
INT008725 | S00226 | U00371 | 2025-04-02T10:00:15Z | train | api | data_analysis | IN | APAC | free | team | internal_policy_qa | openai | gpt-4o-mini | 0.288 | 0.948 | 256 | 156 | 70 | 226 | 1,441 | 0.00452 | 0.861 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | I'm working on something and need help: Can you suggest a few plots to explore churn risk in this dataset of subscriptions? A short, clear answer would be ideal. | Below is a decision-focused breakdown. Goal: handle safety flag rate reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use... | 156.835531 | 2025-04-02 | 10 | 2 | false | true | 2.2286 | Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | 62646e5a7026d4f68ab1867a730451326aa3b664 | 231 | true |
INT008726 | S00651 | U00312 | 2025-04-08T01:24:33Z | train | api | customer_support | US | AMER | free | enterprise_team | support_strict | openai | gpt-4o-mini | 0.372 | 0.923 | 1,024 | 1,014 | 211 | 1,225 | 1,064 | 0.02452 | 0.732 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | high | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Please explain step by step. | Below is a decision-focused breakdown. Goal: handle billing dispute reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 1,151.315789 | 2025-04-08 | 1 | 1 | false | false | 4.8057 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 215 | true |
INT008727 | S01345 | U00315 | 2025-03-29T14:08:09Z | train | web_app | content_writing | BR | AMER | pro | individual | general_assistant | meta | llama-3.1-70b | 0.847 | 0.929 | 1,536 | 311 | 657 | 968 | 1,524 | 0.1936 | 0.576 | true | formatting_error | false | false | false | true | false | false | 0 | none | -1 | negative | false | 2 | regenerate_with_lower_temp | abandoned | high | Context for you: Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. Can you walk me through a good solution? | Let's turn this into a reliable workflow. Signal: formatting_error while working on observability metrics. Quick diagnosis: Your output likely violated the contract. Enforce schema validation... | 635.170604 | 2025-03-29 | 14 | 5 | true | true | 0.4734 | Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | a1c74d04f735d164fdc621e3ce3d5bfea1026243 | 194 | true |
INT008728 | S01274 | U00390 | 2025-02-02T06:12:33Z | val | internal_tool | content_writing | IN | APAC | free | individual | internal_policy_qa | openai | gpt-4.1 | 0.939 | 0.922 | 512 | 906 | 512 | 1,418 | 1,020 | 0.5672 | 0.885 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | abandoned | medium | Quick question: Rewrite this paragraph in a clearer and more professional tone. I'd like a practical, engineering-focused answer. | Below is a practical plan you can execute. Goal: handle A/B testing reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. -... | 1,390.196078 | 2025-02-02 | 6 | 6 | true | false | 1.7695 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | db47a2ba00ac8f8120dcd2fd47d3648a8f594c66 | 222 | true |
INT008729 | S00169 | U00147 | 2025-04-18T09:11:38Z | test | internal_tool | coding_assistant | FR | EMEA | pro | team | analytics_helper | mistral | mistral-large | 0.243 | 0.876 | 1,536 | 172 | 435 | 607 | 988 | 0.07284 | 0.723 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? Can you refactor this function to be more readable and efficient while keeping the same behaviour? | Here's a clear, step-by-step answer. Goal: handle caching strategy reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure... | 614.37247 | 2025-04-18 | 9 | 4 | false | true | 0.3954 | Can you refactor this function to be more readable and efficient while keeping the same behaviour? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | 54f94ac4f6df7ccc4808605ae4fb839b7a1fbfb2 | 210 | true |
INT008730 | S00040 | U00168 | 2025-04-14T14:04:41Z | val | web_app | customer_support | US | AMER | free | team | internal_policy_qa | google | gemini-1.5-pro | 0.1 | 0.92 | 1,024 | 1,050 | 301 | 1,351 | 652 | 0.43232 | 0.803 | false | none | false | false | false | false | false | false | 2 | knowledge_lookup|vector_search | null | no_feedback | false | 0 | none | auto_resolved | low | A user says they were double charged this month. Help me draft a clear explanation and next steps. I'd like a practical, engineering-focused answer. | Below is a decision-focused breakdown. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from... | 2,072.08589 | 2025-04-14 | 14 | 0 | false | true | 3.4884 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 209 | true |
INT008731 | S00157 | U00335 | 2025-04-13T01:40:25Z | train | web_app | customer_support | IN | APAC | enterprise | individual | internal_policy_qa | mistral | mistral-large | 0.384 | 0.893 | 768 | 1,424 | 768 | 2,192 | 534 | 0.26304 | 0.821 | false | none | false | false | false | false | false | false | 3 | code_run|http_request|knowledge_lookup | null | no_feedback | false | 0 | none | auto_resolved | medium | I'm stuck on this and need some guidance: A user says they were double charged this month. Help me draft a clear explanation and next steps. Can you walk me through a good solution? | Let's make the behavior deterministic and observable. Goal: handle login issues reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for login issues and add tests before optimizing. - Add telemetry... | 4,104.868914 | 2025-04-13 | 1 | 6 | true | false | 1.8542 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 262 | true |
INT008732 | S01079 | U00386 | 2025-03-17T15:25:53Z | test | api | brainstorming | IN | APAC | free | individual | analytics_helper | anthropic | claude-3.5-sonnet | 0.623 | 0.838 | 1,536 | 953 | 534 | 1,487 | 719 | 0.47584 | 0.98 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | medium | I'm stuck on this and need some guidance: Give me ten creative ideas for a YouTube series about learning data science from scratch. Can you walk me through a good solution? | This is a common pattern; the fix is straightforward. Goal: handle cost optimization reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for cost optimization and add tests before... | 2,068.150209 | 2025-03-17 | 15 | 0 | false | true | 1.7846 | Give me ten creative ideas for a YouTube series about learning data science from scratch. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | false | 2df52d847c72d1f6b37061123acb53282123086b | 244 | true |
INT008733 | S00070 | U00201 | 2025-03-09T03:56:51Z | train | web_app | customer_support | US | AMER | pro | individual | general_assistant | mistral | mistral-large | 0.807 | 0.953 | 2,048 | 1,084 | 887 | 1,971 | 3,063 | 0.23652 | 0.788 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | Quick question: Our customer is getting an error when trying to log in to their account. Can you help me debug this? Can you walk me through a good solution? | Let's turn this into a reliable workflow. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 643.486778 | 2025-03-09 | 3 | 6 | true | false | 1.2221 | Our customer is getting an error when trying to log in to their account. Can you help me debug this? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | a52fb80193c7c0769c0836852624f4b025a04d2a | 220 | true |
INT008734 | S01129 | U00074 | 2025-04-03T00:03:47Z | val | internal_tool | content_writing | GB | EMEA | enterprise | team | analytics_helper | google | gemini-1.5-pro | 0.479 | 0.887 | 1,536 | 543 | 773 | 1,316 | 390 | 0.42112 | 0.846 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? This is coming from an internal team in our company. Rewrite this paragraph in a clearer and more professional tone. Can you walk me through a good solution? | Let's break this down into actionable steps. Goal: handle A/B testing reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Start by... | 3,374.358974 | 2025-04-03 | 0 | 3 | false | false | 0.7025 | This is coming from an internal team in our company. Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 9716fa3b2c6fe39ac0f75ddaad781dfa757a0de9 | 232 | true |
INT008735 | S01466 | U00020 | 2025-02-05T10:04:16Z | train | web_app | data_analysis | EG | EMEA | free | individual | code_assistant_secure | meta | llama-3.1-70b | 0.243 | 0.866 | 256 | 846 | 72 | 918 | 1,488 | 0.1836 | 0.461 | true | safety_block | false | false | true | false | false | false | 0 | none | null | no_feedback | true | 3 | add_constraints_to_prompt | escalated_to_human | medium | Follow-up: the previous reply did not fully solve the issue. Can you suggest a few plots to explore churn risk in this dataset of subscriptions? A short, clear answer would be ideal. | Here's a clear, step-by-step answer. Signal: safety_block while working on user satisfaction. Quick diagnosis: Log the trigger category to tune your safety policy. Recommended actions: - Add unit tests... | 616.935484 | 2025-02-05 | 10 | 2 | false | true | 11.75 | Follow-up: the previous reply did not fully solve the issue. Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 4e275efe35fdc513983e7d3f3ad32b50cc391969 | 204 | true |
INT008736 | S00705 | U00212 | 2025-03-06T10:58:38Z | train | slack | customer_support | US | AMER | free | individual | support_strict | openai | gpt-4o-mini | 0.925 | 0.852 | 768 | 852 | 179 | 1,031 | 2,881 | 0.02064 | 0.558 | true | hallucination | true | false | false | false | false | false | 0 | none | -2 | negative | true | 0 | regenerate_with_lower_temp | abandoned | high | I'm stuck on this and need some guidance: I need a short, friendly reply to a customer who is asking about the status of their refund. | Let's break this down into actionable steps. Signal: hallucination while working on slow responses. Quick diagnosis: Refuse when confidence is low; avoid confident guesses. Recommended actions: - Introduce... | 357.861854 | 2025-03-06 | 10 | 3 | false | true | 4.7598 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 208 | true |
INT008737 | S00028 | U00372 | 2025-03-17T16:14:21Z | train | api | customer_support | GB | EMEA | free | enterprise_team | support_strict | mistral | mistral-large | 0.481 | 0.895 | 256 | 458 | 256 | 714 | 1,317 | 0.08568 | 0.473 | true | toxicity | false | true | false | false | false | false | 1 | email_send | null | no_feedback | false | 1 | decompose_task | escalated_to_human | medium | Can you help me debug this? Follow-up: the previous reply did not fully solve the issue. A user says they were double charged this month. Help me draft a clear explanation and next steps. Can you walk me through a good solution? | Let's break this down into actionable steps. Signal: toxicity while working on account verification. Quick diagnosis: Measure false positives/negatives so you don't over-block legitimate queries. Recommended... | 542.14123 | 2025-03-17 | 16 | 0 | false | true | 1.7891 | Follow-up: the previous reply did not fully solve the issue. A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 210 | true |
INT008738 | S00490 | U00029 | 2025-02-12T08:11:04Z | train | internal_tool | coding_assistant | FR | EMEA | free | enterprise_team | internal_policy_qa | meta | llama-3.1-8b | 0.058 | 0.963 | 512 | 151 | 194 | 345 | 1,731 | 0.02068 | 0.672 | true | hallucination | true | false | false | false | false | false | 3 | code_run|knowledge_lookup|sql_query | -1 | negative | false | 0 | regenerate_with_lower_temp | escalated_to_human | medium | Context for you: Write a small example that shows how to handle exceptions properly in Go. Can you walk me through a good solution? | Here's a clear, step-by-step answer. Signal: hallucination while working on JSON schema validation. Quick diagnosis: Refuse when confidence is low; avoid confident guesses. Recommended actions:... | 199.306759 | 2025-02-12 | 8 | 2 | false | false | 0.7784 | Write a small example that shows how to handle exceptions properly in Go. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | f84eff1764fee544ced395082209fcd5e43e6a0c | 196 | true |
INT008739 | S00016 | U00227 | 2025-03-06T14:25:29Z | train | internal_tool | brainstorming | GB | EMEA | free | team | creative_writer | google | gemini-1.5-pro | 0.256 | 0.893 | 256 | 445 | 256 | 701 | 781 | 0.22432 | 0.5 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | null | no_feedback | false | 0 | regenerate_with_lower_temp | auto_resolved | high | Can you help me debug this? I need naming ideas for an internal tool that tracks experiments and model runs. A short, clear answer would be ideal. | Here are the guardrails I'd put in place. Signal: tool_error while working on developer productivity. Quick diagnosis: Validate tool arguments before calling the tool; reject invalid calls early.... | 897.567222 | 2025-03-06 | 14 | 3 | false | true | 1.7383 | I need naming ideas for an internal tool that tracks experiments and model runs. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | false | 6f52c3b4cdce68aa614a61f324617ae8d0e91690 | 198 | true |
INT008740 | S01465 | U00158 | 2025-04-29T10:57:30Z | train | web_app | brainstorming | IN | APAC | free | individual | support_strict | mistral | mistral-large | 0.41 | 0.831 | 512 | 1,221 | 214 | 1,435 | 500 | 0.1722 | 0.878 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | high | Quick question: I need naming ideas for an internal tool that tracks experiments and model runs. I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle knowledge base search reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Add telemetry... | 2,870 | 2025-04-29 | 10 | 1 | false | true | 5.7056 | I need naming ideas for an internal tool that tracks experiments and model runs. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | true | e3da6c15e87a6fab71f2d2b735988233f5d1f30b | 245 | true |
INT008741 | S00728 | U00385 | 2025-03-13T13:56:28Z | train | slack | brainstorming | EG | EMEA | free | team | creative_writer | google | gemini-1.5-pro | 0.599 | 0.958 | 768 | 581 | 91 | 672 | 1,204 | 0.21504 | 0.296 | true | hallucination | true | false | false | false | false | false | 0 | none | null | no_feedback | true | 3 | switch_model_family | auto_resolved | high | Context for you: Follow-up: the previous reply did not fully solve the issue. I need naming ideas for an internal tool that tracks experiments and model runs. I'd like a practical, engineering-focused answer. | This is a common pattern; the fix is straightforward. Signal: hallucination while working on LLMOps dashboards. Quick diagnosis: Add grounding: retrieve evidence, require citations, and verify key facts. Recommended... | 558.139535 | 2025-03-13 | 13 | 3 | false | true | 6.3846 | Follow-up: the previous reply did not fully solve the issue. I need naming ideas for an internal tool that tracks experiments and model runs. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | false | 6f52c3b4cdce68aa614a61f324617ae8d0e91690 | 218 | true |
INT008742 | S00211 | U00013 | 2025-03-31T13:33:16Z | test | web_app | content_writing | BR | AMER | pro | enterprise_team | analytics_helper | mistral | mistral-large | 0.746 | 0.758 | 2,048 | 935 | 695 | 1,630 | 2,128 | 0.1956 | 0.764 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | I'm working on something and need help: Rewrite this paragraph in a clearer and more professional tone. I'd like a practical, engineering-focused answer. | Here's the clean way to approach it. Goal: handle tool calling reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Start by writing a small... | 765.977444 | 2025-03-31 | 13 | 0 | false | true | 1.3453 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | 9716fa3b2c6fe39ac0f75ddaad781dfa757a0de9 | 246 | true |
INT008743 | S00185 | U00316 | 2025-03-10T17:01:41Z | test | mobile_app | content_writing | US | AMER | enterprise | individual | creative_writer | openai | gpt-4o-mini | 0.861 | 0.914 | 1,024 | 1,445 | 181 | 1,626 | 670 | 0.03252 | 0.575 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | -2 | negative | false | 0 | switch_model_family | escalated_to_human | critical | I'm working on something and need help: Write a concise product description for a new wireless mouse aimed at software developers. Please explain step by step. | Here's the clean way to approach it. Signal: tool_error while working on prompt injection. Quick diagnosis: Validate tool arguments before calling the tool; reject invalid calls early.... | 2,426.865672 | 2025-03-10 | 17 | 0 | false | true | 7.9834 | Write a concise product description for a new wireless mouse aimed at software developers. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 469eb15f8b08d9b4130e14a6c3888f14e335a854 | 187 | true |
INT008744 | S01140 | U00023 | 2025-03-14T04:32:17Z | train | mobile_app | customer_support | GB | EMEA | free | individual | analytics_helper | local | custom-local-8b | 0.703 | 0.984 | 2,048 | 812 | 239 | 1,051 | 1,653 | 0.0042 | 0.867 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Context for you: Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Please explain step by step. | This is a common pattern; the fix is straightforward. Goal: handle billing dispute reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Prefer simple... | 635.813672 | 2025-03-14 | 4 | 4 | false | false | 3.3975 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 598eba0ae15195966a9835b68c542f0b34fb88e4 | 251 | true |
INT008745 | S00445 | U00031 | 2025-02-23T17:13:37Z | train | web_app | internal_qa | US | AMER | pro | team | creative_writer | meta | llama-3.1-70b | 0.807 | 0.953 | 1,024 | 1,072 | 723 | 1,795 | 2,600 | 0.359 | 0.784 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | abandoned | high | Quick question: Check this draft announcement for any confusing or inconsistent statements. I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle token spikes reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 690.384615 | 2025-02-23 | 17 | 6 | true | true | 1.4827 | Check this draft announcement for any confusing or inconsistent statements. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | false | 3ee501c0eea07616af572d79e74d4f247c4b4851 | 215 | true |
INT008746 | S00730 | U00250 | 2025-04-28T10:10:29Z | val | mobile_app | coding_assistant | IN | APAC | free | individual | support_strict | mistral | mistral-large | 0.216 | 0.847 | 1,024 | 750 | 573 | 1,323 | 1,625 | 0.15876 | 0.625 | true | formatting_error | false | false | false | true | false | false | 1 | http_request | null | no_feedback | false | 0 | regenerate_with_lower_temp | auto_resolved | low | Quick question: I'm getting a JavaScript error: `NullPointerException`. Can you explain what it means and how to fix it? Please explain step by step. | Here's what I'd implement in production. Signal: formatting_error while working on API retry logic. Quick diagnosis: Use constrained generation (schema-first prompt) and add unit tests for formatting.... | 814.153846 | 2025-04-28 | 10 | 0 | false | true | 1.3089 | I'm getting a JavaScript error: `NullPointerException`. Can you explain what it means and how to fix it? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | da34aac46ee8d6d44012043be92e33884c6b4920 | 203 | true |
INT008747 | S00410 | U00449 | 2025-04-26T19:25:44Z | train | web_app | internal_qa | BR | AMER | free | enterprise_team | creative_writer | meta | llama-3.1-8b | 0.727 | 0.879 | 1,024 | 607 | 144 | 751 | 2,111 | 0.04508 | 0.814 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | medium | Explain our refund policy in simple language that a new support agent can remember. I'd like a practical, engineering-focused answer. | Let's break this down into actionable steps. Goal: handle tool failures reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Prefer simple baselines... | 355.755566 | 2025-04-26 | 19 | 5 | true | false | 4.2153 | Explain our refund policy in simple language that a new support agent can remember. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | false | 3ee501c0eea07616af572d79e74d4f247c4b4851 | 254 | true |
INT008748 | S00373 | U00291 | 2025-02-11T00:24:05Z | train | web_app | content_writing | IN | APAC | pro | individual | analytics_helper | mistral | mistral-large | 0.687 | 0.94 | 512 | 365 | 512 | 877 | 680 | 0.10524 | 0.546 | true | formatting_error | false | false | false | true | false | false | 0 | none | null | no_feedback | true | 1 | regenerate_with_lower_temp | auto_resolved | high | Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. | Below is a decision-focused breakdown. Let's stabilize it. Signal: formatting_error while working on tool calling. Quick diagnosis: Your output likely violated the contract. Enforce schema validation and retry-on-invalid. Recommended... | 1,289.705882 | 2025-02-11 | 0 | 1 | false | false | 0.7129 | Follow-up: the previous reply did not fully solve the issue. Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | 9716fa3b2c6fe39ac0f75ddaad781dfa757a0de9 | 236 | true |
INT008749 | S00535 | U00420 | 2025-02-01T12:58:35Z | train | internal_tool | data_analysis | EG | EMEA | free | individual | internal_policy_qa | meta | llama-3.1-70b | 0.256 | 0.93 | 2,048 | 1,213 | 116 | 1,329 | 854 | 0.2658 | 0.98 | false | none | false | false | false | false | false | false | 2 | http_request|knowledge_lookup | null | no_feedback | false | 0 | none | auto_resolved | high | Can you help me debug this? Help me compute the conversion rate per marketing channel and highlight the top 3 performers. Can you walk me through a good solution? | Below is a decision-focused breakdown. Goal: handle error rate reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Add telemetry fields for... | 1,556.206089 | 2025-02-01 | 12 | 5 | true | true | 10.4569 | Help me compute the conversion rate per marketing channel and highlight the top 3 performers. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 62646e5a7026d4f68ab1867a730451326aa3b664 | 241 | true |
INT008750 | S00029 | U00342 | 2025-04-19T10:09:02Z | train | internal_tool | data_analysis | DE | EMEA | free | enterprise_team | support_strict | openai | gpt-4.1 | 0.27 | 0.918 | 256 | 1,298 | 256 | 1,554 | 1,409 | 0.6216 | 0.94 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | escalated_to_human | high | Context for you: Can you suggest a few plots to explore churn risk in this dataset of subscriptions? Can you walk me through a good solution? | Here's what I'd implement in production. Goal: handle latency p95 reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use schema validation... | 1,102.909865 | 2025-04-19 | 10 | 5 | true | true | 5.0703 | Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | a15ac130f4f428f0efac738d5c68655674fccc6e | 246 | true |
INT008751 | S00903 | U00291 | 2025-02-24T22:33:24Z | train | api | data_analysis | IN | APAC | pro | individual | creative_writer | google | gemini-1.5-pro | 0.214 | 0.858 | 512 | 1,491 | 14 | 1,505 | 1,106 | 0.4816 | 0.877 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | I'm stuck on this and need some guidance: I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. | Here's the clean way to approach it. Goal: handle user satisfaction reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure... | 1,360.759494 | 2025-02-24 | 22 | 0 | false | false | 106.5 | I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 814335a175319858ca540d38bfcac020ec643c3f | 211 | true |
INT008752 | S01099 | U00358 | 2025-02-22T10:20:55Z | train | web_app | content_writing | FR | EMEA | pro | individual | code_assistant_secure | anthropic | claude-3.5-sonnet | 1.037 | 0.893 | 1,536 | 444 | 448 | 892 | 944 | 0.28544 | 0.976 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | abandoned | low | Rewrite this paragraph in a clearer and more professional tone. A short, clear answer would be ideal. | This is a common pattern; the fix is straightforward. Goal: handle observability metrics reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug... | 944.915254 | 2025-02-22 | 10 | 5 | true | true | 0.9911 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | 20f1dad1d39f7b8df6faaa166fe78024f8d03fd6 | 224 | true |
INT008753 | S01126 | U00321 | 2025-04-18T13:27:40Z | train | api | customer_support | EG | EMEA | pro | enterprise_team | general_assistant | anthropic | claude-3.5-sonnet | 0.943 | 0.853 | 768 | 618 | 426 | 1,044 | 405 | 0.33408 | 0.7 | false | none | false | false | false | false | false | false | 2 | email_send|sql_query | 2 | positive | false | 0 | none | auto_resolved | medium | Context for you: Our customer is getting an error when trying to log in to their account. Can you help me debug this? Can you walk me through a good solution? | Here's a clear, step-by-step answer. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Start by... | 2,577.777778 | 2025-04-18 | 13 | 4 | false | true | 1.4507 | Our customer is getting an error when trying to log in to their account. Can you help me debug this? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | a52fb80193c7c0769c0836852624f4b025a04d2a | 235 | true |
INT008754 | S00323 | U00290 | 2025-03-21T10:47:36Z | test | slack | coding_assistant | DE | EMEA | free | team | general_assistant | local | custom-local-8b | 0.385 | 0.86 | 512 | 242 | 200 | 442 | 1,412 | 0.00176 | 0.73 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | medium | I'm working on something and need help: I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. Please explain step by step. | Here are the guardrails I'd put in place. Goal: handle rate limiting reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 313.031161 | 2025-03-21 | 10 | 4 | false | true | 1.21 | I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | 3ddc6acb5986d68a027a4c760a5adfbe894b8deb | 216 | true |
INT008755 | S00165 | U00089 | 2025-02-09T18:59:44Z | train | api | customer_support | BR | AMER | pro | team | support_strict | local | custom-local-8b | 0.84 | 0.862 | 256 | 983 | 183 | 1,166 | 500 | 0.00468 | 0.65 | false | none | false | false | false | false | false | false | 1 | knowledge_lookup | null | no_feedback | false | 0 | none | auto_resolved | low | Quick question: Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use schema... | 2,332 | 2025-02-09 | 18 | 6 | true | true | 5.3716 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 242 | true |
INT008756 | S00934 | U00302 | 2025-02-06T19:09:56Z | train | api | coding_assistant | EG | EMEA | free | team | creative_writer | openai | gpt-4o-mini | 0.233 | 0.887 | 1,536 | 422 | 430 | 852 | 859 | 0.01704 | 0.767 | false | none | false | false | false | false | false | false | 3 | code_run|sql_query|workflow_orchestrator | null | no_feedback | false | 0 | none | auto_resolved | low | I'm stuck on this and need some guidance: I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. I'd like a practical, engineering-focused answer. | Here's a clear, step-by-step answer. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for async I/O pitfalls and add tests before optimizing. -... | 991.85099 | 2025-02-06 | 19 | 3 | false | false | 0.9814 | I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | 83c939e9b891e505ac06548e158556e919317a64 | 243 | true |
INT008757 | S01592 | U00252 | 2025-02-07T01:09:45Z | train | web_app | internal_qa | IN | APAC | enterprise | team | general_assistant | openai | gpt-4o-mini | 0.401 | 0.827 | 1,536 | 1,345 | 163 | 1,508 | 953 | 0.03016 | 0.464 | true | toxicity | false | true | false | false | false | false | 0 | none | null | no_feedback | true | 3 | decompose_task | auto_resolved | high | I'm working on something and need help: Follow-up: the previous reply did not fully solve the issue. Check this draft announcement for any confusing or inconsistent statements. Please explain step by step. | Here's what I'd implement in production. Signal: toxicity while working on latency. Quick diagnosis: Measure false positives/negatives so you don't over-block legitimate queries. Recommended actions: - Add... | 1,582.371459 | 2025-02-07 | 1 | 4 | false | false | 8.2515 | Follow-up: the previous reply did not fully solve the issue. Check this draft announcement for any confusing or inconsistent statements. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | false | 7cd4f5f09ad56cc7a4cac3fea1f795d4e2ccc742 | 208 | true |
INT008758 | S00682 | U00215 | 2025-04-18T18:42:46Z | val | mobile_app | data_analysis | EG | EMEA | free | enterprise_team | creative_writer | openai | gpt-4.1 | 0.281 | 0.961 | 256 | 750 | 256 | 1,006 | 1,365 | 0.4024 | 0.694 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Quick question: I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. Please explain step by step. | I'll keep this concrete and measurable. Goal: handle tool-call success rate reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use... | 736.996337 | 2025-04-18 | 18 | 4 | false | true | 2.9297 | I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | 814335a175319858ca540d38bfcac020ec643c3f | 233 | true |
INT008759 | S01301 | U00418 | 2025-04-15T00:46:40Z | train | web_app | customer_support | IN | APAC | pro | team | general_assistant | local | custom-local-8b | 0.489 | 0.88 | 512 | 187 | 512 | 699 | 1,753 | 0.0028 | 0.887 | false | none | false | false | false | false | false | false | 3 | email_send|http_request|knowledge_lookup | null | no_feedback | false | 0 | none | auto_resolved | critical | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Can you walk me through a good solution? | Below is a decision-focused breakdown. Goal: handle data export request reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Prefer simple... | 398.745009 | 2025-04-15 | 0 | 1 | false | false | 0.3652 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | a52fb80193c7c0769c0836852624f4b025a04d2a | 244 | true |
INT008760 | S00599 | U00013 | 2025-03-07T22:21:09Z | train | slack | customer_support | BR | AMER | pro | enterprise_team | general_assistant | local | custom-local-8b | 0.615 | 0.909 | 1,024 | 1,072 | 901 | 1,973 | 954 | 0.00788 | 0.565 | true | formatting_error | false | false | false | true | false | false | 0 | none | -1 | negative | false | 1 | manual_post_edit | auto_resolved | high | Context for you: Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Can you walk me through a good solution? | Let's turn this into a reliable workflow. Signal: formatting_error while working on billing dispute. Quick diagnosis: Fail closed: if invalid, don't pass downstream; retry with the validator error.... | 2,068.134172 | 2025-03-07 | 22 | 4 | false | false | 1.1898 | Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | a52fb80193c7c0769c0836852624f4b025a04d2a | 200 | true |
INT008761 | S00941 | U00352 | 2025-02-23T11:27:45Z | train | api | data_analysis | FR | EMEA | enterprise | individual | creative_writer | anthropic | claude-3.5-sonnet | 0.316 | 0.881 | 768 | 1,348 | 768 | 2,116 | 927 | 0.67712 | 0.717 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | Context for you: Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | Let's turn this into a reliable workflow. Goal: handle latency p95 reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Add telemetry fields for... | 2,282.632147 | 2025-02-23 | 11 | 6 | true | true | 1.7552 | Can you suggest a few plots to explore churn risk in this dataset of subscriptions? | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | 814335a175319858ca540d38bfcac020ec643c3f | 246 | true |
INT008762 | S00395 | U00061 | 2025-03-11T12:49:15Z | train | web_app | coding_assistant | EG | EMEA | pro | enterprise_team | analytics_helper | google | gemini-1.5-pro | 0.454 | 0.871 | 256 | 390 | 256 | 646 | 1,809 | 0.20672 | 0.712 | false | none | false | false | false | false | false | false | 1 | sql_query | null | no_feedback | false | 0 | none | abandoned | medium | I'm getting a Java error: `TypeError`. Can you explain what it means and how to fix it? Can you walk me through a good solution? | Here's a clear, step-by-step answer. Goal: handle JSON schema validation reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. -... | 357.103372 | 2025-03-11 | 12 | 1 | false | true | 1.5234 | I'm getting a Java error: `TypeError`. Can you explain what it means and how to fix it? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | 54f94ac4f6df7ccc4808605ae4fb839b7a1fbfb2 | 227 | true |
INT008763 | S00404 | U00262 | 2025-03-28T09:27:33Z | train | api | content_writing | IN | APAC | free | enterprise_team | code_assistant_secure | google | gemini-1.5-pro | 0.814 | 0.889 | 512 | 618 | 395 | 1,013 | 986 | 0.32416 | 0.864 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | medium | Turn these bullet points into a short, engaging blog introduction about learning machine learning. Please explain step by step. | Here's a clear, step-by-step answer. Goal: handle A/B testing reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. -... | 1,027.383367 | 2025-03-28 | 9 | 4 | false | true | 1.5646 | Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | 20f1dad1d39f7b8df6faaa166fe78024f8d03fd6 | 215 | true |
INT008764 | S00179 | U00449 | 2025-03-27T23:14:54Z | train | slack | customer_support | BR | AMER | free | enterprise_team | internal_policy_qa | openai | gpt-4o-mini | 0.1 | 0.876 | 2,048 | 1,114 | 616 | 1,730 | 926 | 0.0346 | 0.498 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | 0 | neutral | true | 2 | add_constraints_to_prompt | auto_resolved | low | Context for you: Follow-up: the previous reply did not fully solve the issue. A user says they were double charged this month. Help me draft a clear explanation and next steps. Please explain step by step. | Below is a practical plan you can execute. Signal: tool_error while working on unexpected charge. Quick diagnosis: Separate tool telemetry from model telemetry so you can see which component failed.... | 1,868.25054 | 2025-03-27 | 23 | 3 | false | false | 1.8084 | Follow-up: the previous reply did not fully solve the issue. A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 03bf57d7adc83026d525c8192379aa203b71e1eb | 201 | true |
INT008765 | S01468 | U00290 | 2025-02-11T20:52:56Z | train | web_app | customer_support | DE | EMEA | free | team | internal_policy_qa | local | custom-local-8b | 0.406 | 0.926 | 512 | 248 | 345 | 593 | 1,949 | 0.00236 | 0.638 | false | none | false | false | false | false | false | false | 2 | email_send|http_request | null | no_feedback | false | 0 | none | auto_resolved | medium | I'm stuck on this and need some guidance: I need a short, friendly reply to a customer who is asking about the status of their refund. Please explain step by step. | Here are the guardrails I'd put in place. Goal: handle missing feature reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Start by... | 304.258594 | 2025-02-11 | 20 | 1 | false | false | 0.7188 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 233 | true |
INT008766 | S01101 | U00344 | 2025-04-19T02:41:52Z | train | api | customer_support | US | AMER | free | individual | general_assistant | meta | llama-3.1-70b | 0.717 | 0.978 | 512 | 362 | 497 | 859 | 1,214 | 0.1718 | 0.98 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | high | Context for you: Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Can you walk me through a good solution? | This is a common pattern; the fix is straightforward. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use... | 707.578254 | 2025-04-19 | 2 | 5 | true | false | 0.7284 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | a52fb80193c7c0769c0836852624f4b025a04d2a | 242 | true |
INT008767 | S00522 | U00154 | 2025-03-25T11:35:30Z | train | api | data_analysis | GB | EMEA | free | individual | general_assistant | meta | llama-3.1-70b | 0.153 | 0.862 | 1,024 | 642 | 305 | 947 | 1,189 | 0.1894 | 0.898 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | critical | Context for you: I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle latency p95 reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use schema... | 796.46762 | 2025-03-25 | 11 | 1 | false | true | 2.1049 | I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 0af51e7e9a94eeacb4db5e80369428b0b639dc4e | 231 | true |
INT008768 | S00315 | U00048 | 2025-03-16T16:33:16Z | train | web_app | customer_support | IN | APAC | free | enterprise_team | creative_writer | openai | gpt-4.1 | 0.525 | 0.865 | 512 | 1,179 | 426 | 1,605 | 2,112 | 0.642 | 0.927 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle billing dispute reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Prefer simple... | 759.943182 | 2025-03-16 | 16 | 6 | true | true | 2.7676 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 7a490bfafd7436aea43a528e9fe4cb6db10b9a6b | 243 | true |
INT008769 | S00281 | U00169 | 2025-04-04T13:41:17Z | train | api | data_analysis | US | AMER | pro | individual | analytics_helper | google | gemini-1.5-pro | 0.336 | 0.941 | 1,024 | 1,006 | 638 | 1,644 | 2,056 | 0.52608 | 0.834 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Can you help me debug this? Help me compute the conversion rate per marketing channel and highlight the top 3 performers. Please explain step by step. | Below is a decision-focused breakdown. Goal: handle user satisfaction reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use schema... | 799.610895 | 2025-04-04 | 13 | 4 | false | true | 1.5768 | Help me compute the conversion rate per marketing channel and highlight the top 3 performers. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | fbaac57cdb64395ff60fbf79b148e22caaff1431 | 234 | true |
INT008770 | S01454 | U00200 | 2025-04-11T12:40:12Z | train | web_app | data_analysis | GB | EMEA | free | enterprise_team | creative_writer | meta | llama-3.1-70b | 0.159 | 0.925 | 1,024 | 1,007 | 662 | 1,669 | 1,371 | 0.3338 | 0.98 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | medium | Context for you: Help me compute the conversion rate per marketing channel and highlight the top 3 performers. I'd like a practical, engineering-focused answer. | Below is a practical plan you can execute. Goal: handle tool-call success rate reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for tool-call success rate and add tests before... | 1,217.359592 | 2025-04-11 | 12 | 4 | false | true | 1.5211 | Help me compute the conversion rate per marketing channel and highlight the top 3 performers. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | false | 814335a175319858ca540d38bfcac020ec643c3f | 243 | true |
INT008771 | S01476 | U00068 | 2025-04-08T22:43:10Z | train | slack | content_writing | CA | AMER | pro | individual | general_assistant | openai | gpt-4o-mini | 0.852 | 0.931 | 1,536 | 1,485 | 164 | 1,649 | 708 | 0.033 | 0.813 | false | none | false | false | false | false | false | false | 0 | none | 0 | neutral | false | 0 | none | abandoned | low | I'm stuck on this and need some guidance: Rewrite this paragraph in a clearer and more professional tone. Can you walk me through a good solution? | We can make this robust with a few changes. Goal: handle tool calling reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Prefer... | 2,329.096045 | 2025-04-08 | 22 | 1 | false | false | 9.0549 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | a1c74d04f735d164fdc621e3ce3d5bfea1026243 | 235 | true |
INT008772 | S01048 | U00149 | 2025-04-19T22:23:01Z | test | web_app | brainstorming | CA | AMER | pro | team | analytics_helper | meta | llama-3.1-70b | 0.629 | 0.972 | 1,024 | 487 | 485 | 972 | 870 | 0.1944 | 0.669 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? Brainstorm taglines for a productivity app that helps engineers stay focused. I'd like a practical, engineering-focused answer. | Let's break this down into actionable steps. Goal: handle developer productivity reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Start by... | 1,117.241379 | 2025-04-19 | 22 | 5 | true | false | 1.0041 | Brainstorm taglines for a productivity app that helps engineers stay focused. | You are a creative assistant helping users explore ideas and options. Suggest multiple alternatives, structure them clearly, and stay practical. | true | 2df52d847c72d1f6b37061123acb53282123086b | 244 | true |
INT008773 | S00891 | U00181 | 2025-02-07T23:27:40Z | train | web_app | customer_support | BR | AMER | enterprise | individual | support_strict | mistral | mistral-large | 0.732 | 0.924 | 768 | 1,358 | 50 | 1,408 | 1,566 | 0.16896 | 0.558 | true | tool_error | false | false | false | false | true | false | 1 | tool_invocation_failed | -1 | negative | false | 0 | enable_tools | auto_resolved | low | Context for you: Our customer is getting an error when trying to log in to their account. Can you help me debug this? Please explain step by step. | Below is a practical plan you can execute. Let's stabilize it. Signal: tool_error while working on missing feature. Quick diagnosis: Validate tool arguments before calling the tool; reject invalid calls early. Recommended actions: - Add... | 899.106003 | 2025-02-07 | 23 | 4 | false | false | 27.16 | Our customer is getting an error when trying to log in to their account. Can you help me debug this? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 239 | true |
INT008774 | S01147 | U00208 | 2025-03-07T17:55:25Z | train | slack | data_analysis | US | AMER | free | individual | support_strict | meta | llama-3.1-70b | 0.412 | 0.885 | 1,024 | 992 | 636 | 1,628 | 1,887 | 0.3256 | 0.643 | true | formatting_error | false | false | false | true | false | false | 0 | none | -1 | negative | true | 1 | manual_post_edit | auto_resolved | low | I'm stuck on this and need some guidance: Follow-up: the previous reply did not fully solve the issue. Help me compute the conversion rate per marketing channel and highlight the top 3 performers. Can you walk me through a good solution? | Let's make the behavior deterministic and observable. Signal: formatting_error while working on latency p95. Quick diagnosis: Use constrained generation (schema-first prompt) and add unit tests for formatting.... | 862.745098 | 2025-03-07 | 17 | 4 | false | true | 1.5597 | Follow-up: the previous reply did not fully solve the issue. Help me compute the conversion rate per marketing channel and highlight the top 3 performers. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | a15ac130f4f428f0efac738d5c68655674fccc6e | 212 | true |
INT008775 | S01012 | U00449 | 2025-03-25T22:08:39Z | train | slack | customer_support | BR | AMER | free | enterprise_team | code_assistant_secure | mistral | mistral-large | 0.475 | 0.819 | 2,048 | 250 | 55 | 305 | 949 | 0.0366 | 0.672 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | abandoned | medium | I need a short, friendly reply to a customer who is asking about the status of their refund. I'd like a practical, engineering-focused answer. | Let's make the behavior deterministic and observable. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from... | 321.390938 | 2025-03-25 | 22 | 1 | false | false | 4.5455 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 1a8e13c1d6c8047debb9a230bf3ae08524ccb884 | 224 | true |
INT008776 | S01025 | U00343 | 2025-04-15T21:43:35Z | train | web_app | content_writing | US | AMER | enterprise | individual | internal_policy_qa | meta | llama-3.1-8b | 0.815 | 0.899 | 768 | 1,102 | 768 | 1,870 | 1,348 | 0.1122 | 0.925 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | abandoned | low | I'm stuck on this and need some guidance: Turn these bullet points into a short, engaging blog introduction about learning machine learning. I'd like a practical, engineering-focused answer. | Let's break this down into actionable steps. Goal: handle tool calling reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Use schema validation (e.g.,... | 1,387.240356 | 2025-04-15 | 21 | 1 | false | false | 1.4349 | Turn these bullet points into a short, engaging blog introduction about learning machine learning. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | db47a2ba00ac8f8120dcd2fd47d3648a8f594c66 | 253 | true |
INT008777 | S00469 | U00161 | 2025-04-07T12:00:03Z | test | api | data_analysis | FR | EMEA | free | individual | general_assistant | meta | llama-3.1-8b | 0.434 | 0.955 | 768 | 125 | 427 | 552 | 1,503 | 0.03312 | 0.763 | false | none | false | false | false | false | false | false | 3 | sql_query|vector_search|workflow_orchestrator | 2 | positive | false | 0 | none | auto_resolved | high | Context for you: I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. Please explain step by step. | Below is a practical plan you can execute. Goal: handle tool-call success rate reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Start by... | 367.265469 | 2025-04-07 | 12 | 0 | false | true | 0.2927 | I have a CSV with user sessions. Show me a quick breakdown of sessions per country and device type. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | 0af51e7e9a94eeacb4db5e80369428b0b639dc4e | 246 | true |
INT008778 | S00259 | U00218 | 2025-04-22T21:26:40Z | train | web_app | customer_support | US | AMER | enterprise | individual | code_assistant_secure | google | gemini-1.5-pro | 0.814 | 0.968 | 1,024 | 1,035 | 908 | 1,943 | 642 | 0.62176 | 0.647 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | I'm working on something and need help: A user says they were double charged this month. Help me draft a clear explanation and next steps. A short, clear answer would be ideal. | Below is a decision-focused breakdown. Goal: handle billing dispute reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 3,026.479751 | 2025-04-22 | 21 | 1 | false | false | 1.1399 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 1a8e13c1d6c8047debb9a230bf3ae08524ccb884 | 215 | true |
INT008779 | S00790 | U00440 | 2025-04-09T15:15:07Z | test | slack | internal_qa | BR | AMER | free | team | internal_policy_qa | openai | gpt-4o-mini | 0.327 | 0.931 | 2,048 | 768 | 314 | 1,082 | 1,459 | 0.02164 | 0.791 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Can you help me debug this? Explain our refund policy in simple language that a new support agent can remember. A short, clear answer would be ideal. | I'll keep this concrete and measurable. Goal: handle safety flags reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 741.603838 | 2025-04-09 | 15 | 2 | false | true | 2.4459 | Explain our refund policy in simple language that a new support agent can remember. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | false | 8ad879fd7895b38236ec6ff25a956251dc06e51b | 213 | true |
INT008780 | S00220 | U00335 | 2025-03-07T03:49:57Z | train | api | customer_support | IN | APAC | enterprise | individual | internal_policy_qa | meta | llama-3.1-8b | 0.732 | 0.899 | 2,048 | 215 | 789 | 1,004 | 1,014 | 0.06024 | 0.927 | false | none | false | false | false | false | false | false | 2 | email_send|workflow_orchestrator | 1 | positive | false | 1 | regenerate_with_lower_temp | auto_resolved | medium | I'm stuck on this and need some guidance: Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle slow responses reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices.... | 990.138067 | 2025-03-07 | 3 | 4 | false | false | 0.2725 | Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 03bf57d7adc83026d525c8192379aa203b71e1eb | 221 | true |
INT008781 | S01254 | U00250 | 2025-02-04T12:42:36Z | train | api | customer_support | IN | APAC | free | individual | internal_policy_qa | openai | gpt-4o-mini | 0.622 | 0.959 | 1,024 | 916 | 567 | 1,483 | 1,174 | 0.02968 | 0.45 | true | toxicity | false | true | false | false | false | false | 0 | none | null | no_feedback | true | 0 | enable_tools | auto_resolved | medium | I'm working on something and need help: Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Can you walk me through a good solution? | Below is a decision-focused breakdown. Signal: toxicity while working on account verification. Quick diagnosis: Add a rewrite pathway: sanitize unsafe content into safe alternatives.... | 1,263.202726 | 2025-02-04 | 12 | 1 | false | true | 1.6155 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 185 | true |
INT008782 | S00547 | U00426 | 2025-04-24T10:21:18Z | train | web_app | coding_assistant | FR | EMEA | free | team | support_strict | anthropic | claude-3.5-sonnet | 0.327 | 0.807 | 1,024 | 240 | 417 | 657 | 1,976 | 0.21024 | 0.866 | false | none | false | false | false | false | false | false | 2 | knowledge_lookup|workflow_orchestrator | null | no_feedback | false | 0 | none | auto_resolved | low | Quick question: I'm getting a JavaScript error: `TypeError`. Can you explain what it means and how to fix it? Can you walk me through a good solution? | Here's what I'd implement in production. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions... | 332.489879 | 2025-04-24 | 10 | 3 | false | true | 0.5755 | I'm getting a JavaScript error: `TypeError`. Can you explain what it means and how to fix it? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | da34aac46ee8d6d44012043be92e33884c6b4920 | 220 | true |
INT008783 | S00345 | U00076 | 2025-04-30T09:59:04Z | train | web_app | content_writing | IN | APAC | free | enterprise_team | support_strict | google | gemini-1.5-pro | 0.616 | 0.817 | 2,048 | 648 | 44 | 692 | 2,737 | 0.22144 | 0.756 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | medium | I'm working on something and need help: Write a concise product description for a new wireless mouse aimed at software developers. A short, clear answer would be ideal. | Here's a clear, step-by-step answer. Goal: handle A/B testing reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. -... | 252.831567 | 2025-04-30 | 9 | 2 | false | true | 14.7273 | Write a concise product description for a new wireless mouse aimed at software developers. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 75b7366387b6cf4f5631ec9dd298a359e5bb9c1d | 216 | true |
INT008784 | S00908 | U00209 | 2025-02-25T03:18:00Z | val | web_app | coding_assistant | IN | APAC | pro | individual | analytics_helper | meta | llama-3.1-8b | 0.23 | 0.934 | 1,536 | 517 | 811 | 1,328 | 2,654 | 0.07968 | 0.787 | false | none | false | false | false | false | false | false | 1 | knowledge_lookup | null | no_feedback | false | 0 | none | auto_resolved | low | Quick question: Can you refactor this function to be more readable and efficient while keeping the same behaviour? Can you walk me through a good solution? | Here's what I'd implement in production. Goal: handle JSON schema validation reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly.... | 500.37679 | 2025-02-25 | 3 | 1 | false | false | 0.6375 | Can you refactor this function to be more readable and efficient while keeping the same behaviour? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | 54f94ac4f6df7ccc4808605ae4fb839b7a1fbfb2 | 233 | true |
INT008785 | S00328 | U00313 | 2025-03-11T01:23:26Z | train | web_app | internal_qa | IN | APAC | enterprise | enterprise_team | support_strict | meta | llama-3.1-8b | 0.307 | 0.947 | 1,536 | 949 | 797 | 1,746 | 1,602 | 0.10476 | 0.765 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | Context for you: Explain our refund policy in simple language that a new support agent can remember. | We can make this robust with a few changes. Goal: handle latency reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use schema validation... | 1,089.88764 | 2025-03-11 | 1 | 1 | false | false | 1.1907 | Explain our refund policy in simple language that a new support agent can remember. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | true | 4e3bb3b7c597daf93de3f611632f4c56cefdec0e | 245 | true |
INT008786 | S01222 | U00156 | 2025-04-21T17:11:02Z | train | slack | internal_qa | US | AMER | free | team | analytics_helper | meta | llama-3.1-8b | 0.367 | 0.939 | 1,024 | 658 | 163 | 821 | 672 | 0.04924 | 0.531 | true | hallucination | true | false | false | false | false | false | 0 | none | -2 | negative | true | 0 | switch_model_family | escalated_to_human | medium | Can you help me debug this? Summarise this internal security policy into three bullet points for non-technical employees. Please explain step by step. | Here's a clear, step-by-step answer. Signal: hallucination while working on token spikes. Quick diagnosis: Use a 'claim -> evidence -> answer' structure and log missing evidence. Recommended actions: -... | 1,221.72619 | 2025-04-21 | 17 | 0 | false | true | 4.0368 | Summarise this internal security policy into three bullet points for non-technical employees. | You are an internal knowledge assistant helping teammates navigate documentation, policies, and internal tools. Be precise, concise, and point to relevant information and next steps. | true | d33eca392020f2e6e14ec7fe3091a25b46d6bae1 | 204 | true |
INT008787 | S00821 | U00192 | 2025-04-06T09:07:12Z | train | web_app | coding_assistant | AU | APAC | free | enterprise_team | analytics_helper | anthropic | claude-3.5-sonnet | 0.193 | 0.872 | 1,024 | 120 | 465 | 585 | 655 | 0.1872 | 0.599 | true | hallucination | true | false | false | false | false | false | 3 | code_run|vector_search|workflow_orchestrator | null | no_feedback | false | 1 | regenerate_with_lower_temp | abandoned | low | I'm working on something and need help: Follow-up: the previous reply did not fully solve the issue. I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. Please explain step by step. | Below is a practical plan you can execute. Signal: hallucination while working on API retry logic. Quick diagnosis: Add grounding: retrieve evidence, require citations, and verify key facts.... | 893.129771 | 2025-04-06 | 9 | 6 | true | true | 0.2581 | Follow-up: the previous reply did not fully solve the issue. I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | 54f94ac4f6df7ccc4808605ae4fb839b7a1fbfb2 | 193 | true |
INT008788 | S01075 | U00093 | 2025-04-17T13:41:18Z | train | api | customer_support | US | AMER | pro | individual | code_assistant_secure | openai | gpt-4o-mini | 1.1 | 0.976 | 768 | 259 | 78 | 337 | 820 | 0.00676 | 0.662 | false | none | false | false | false | false | false | false | 3 | code_run|email_send|http_request | 1 | positive | false | 1 | regenerate_with_lower_temp | abandoned | medium | Context for you: Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? Please explain step by step. | We can make this robust with a few changes. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure slices. - Start by... | 410.97561 | 2025-04-17 | 13 | 3 | false | true | 3.3205 | Follow-up: the previous reply did not fully solve the issue. Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | 1a8e13c1d6c8047debb9a230bf3ae08524ccb884 | 237 | true |
INT008789 | S00247 | U00129 | 2025-04-21T07:17:23Z | train | api | coding_assistant | FR | EMEA | free | individual | support_strict | google | gemini-1.5-pro | 0.156 | 0.986 | 1,536 | 80 | 369 | 449 | 1,849 | 0.14368 | 0.88 | false | none | false | false | false | false | false | false | 3 | code_run|http_request|sql_query | null | no_feedback | false | 0 | none | auto_resolved | medium | Can you help me debug this? I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. I'd like a practical, engineering-focused answer. | Let's turn this into a reliable workflow. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use schema... | 242.833964 | 2025-04-21 | 7 | 0 | false | false | 0.2168 | I need help understanding why this asynchronous code is not awaiting correctly. Explain with a small code snippet. | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | da34aac46ee8d6d44012043be92e33884c6b4920 | 243 | true |
INT008790 | S00708 | U00125 | 2025-03-01T10:38:07Z | train | internal_tool | content_writing | AE | EMEA | pro | individual | general_assistant | mistral | mistral-large | 0.91 | 0.927 | 512 | 318 | 512 | 830 | 1,348 | 0.0996 | 0.619 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | I'm stuck on this and need some guidance: Rewrite this paragraph in a clearer and more professional tone. A short, clear answer would be ideal. | I'll keep this concrete and measurable. Goal: handle tool calling reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Use... | 615.727003 | 2025-03-01 | 10 | 5 | true | true | 0.6211 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | a1c74d04f735d164fdc621e3ce3d5bfea1026243 | 228 | true |
INT008791 | S01198 | U00319 | 2025-03-16T07:35:21Z | val | api | customer_support | DE | EMEA | pro | individual | internal_policy_qa | openai | gpt-4o-mini | 0.734 | 0.901 | 1,024 | 1,346 | 459 | 1,805 | 811 | 0.03608 | 0.869 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | I'll keep this concrete and measurable. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Prefer simple baselines... | 2,225.647349 | 2025-03-16 | 7 | 6 | true | false | 2.9325 | Hi, I'm having an issue with my billing. The invoice total doesn't match what I expected. Can you check it? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 249 | true |
INT008792 | S01168 | U00115 | 2025-02-15T17:29:20Z | train | web_app | coding_assistant | US | AMER | free | individual | internal_policy_qa | mistral | mistral-large | 0.211 | 0.887 | 768 | 1,343 | 211 | 1,554 | 2,067 | 0.18648 | 0.747 | false | none | false | false | false | false | false | false | 1 | http_request | 0 | neutral | false | 0 | none | auto_resolved | low | Quick question: I'm getting a Java error: `IndexError`. Can you explain what it means and how to fix it? Can you walk me through a good solution? | This is a common pattern; the fix is straightforward. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for async I/O pitfalls and add... | 751.814224 | 2025-02-15 | 17 | 5 | true | true | 6.3649 | I'm getting a Java error: `IndexError`. Can you explain what it means and how to fix it? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | true | f84eff1764fee544ced395082209fcd5e43e6a0c | 233 | true |
INT008793 | S01581 | U00306 | 2025-04-15T00:04:42Z | test | api | data_analysis | DE | EMEA | pro | team | support_strict | google | gemini-1.5-pro | 0.16 | 0.87 | 512 | 1,257 | 471 | 1,728 | 1,024 | 0.55296 | 0.815 | false | none | false | false | false | false | false | false | 0 | none | 0 | neutral | false | 0 | none | auto_resolved | medium | Context for you: I want a short summary of key metrics for this dataset: total users, active users, and average revenue per user. A short, clear answer would be ideal. | Here's the clean way to approach it. Goal: handle safety flag rate reliably, with guardrails for formatting and failure handling. Plan: - Prefer simple baselines first; then iterate based on measured failure... | 1,687.5 | 2025-04-15 | 0 | 1 | false | false | 2.6688 | I want a short summary of key metrics for this dataset: total users, active users, and average revenue per user. | You are a data analysis assistant helping users reason about data, metrics, and models. Explain trade-offs, outline steps, and keep the language accessible but precise. | true | a15ac130f4f428f0efac738d5c68655674fccc6e | 210 | true |
INT008794 | S01363 | U00253 | 2025-03-11T10:50:13Z | test | internal_tool | coding_assistant | IN | APAC | free | individual | internal_policy_qa | mistral | mistral-large | 0.146 | 0.873 | 512 | 240 | 433 | 673 | 1,000 | 0.08076 | 0.93 | false | none | false | false | false | false | false | false | 1 | code_run | null | no_feedback | false | 0 | none | auto_resolved | medium | Quick question: Can you refactor this function to be more readable and efficient while keeping the same behaviour? | Below is a decision-focused breakdown. Goal: handle async I/O pitfalls reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for async I/O pitfalls and add... | 673 | 2025-03-11 | 10 | 1 | false | true | 0.5543 | Can you refactor this function to be more readable and efficient while keeping the same behaviour? | You are a senior software engineer helping users with code, debugging, and best practices. Provide clear explanations, structured steps, and high-quality code examples when relevant. | false | f84eff1764fee544ced395082209fcd5e43e6a0c | 218 | true |
INT008795 | S01196 | U00074 | 2025-04-08T10:56:33Z | train | web_app | content_writing | GB | EMEA | enterprise | team | general_assistant | mistral | mistral-large | 0.66 | 0.914 | 768 | 1,363 | 395 | 1,758 | 894 | 0.21096 | 0.89 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | Can you help me debug this? Draft a friendly email announcing a small price increase to our existing customers. Please explain step by step. | We can make this robust with a few changes. Goal: handle evaluation for LLM systems reliably, with guardrails for formatting and failure handling. Plan: - Start by writing a small function boundary for evaluation for LLM... | 1,966.442953 | 2025-04-08 | 10 | 1 | false | true | 3.4506 | Draft a friendly email announcing a small price increase to our existing customers. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | false | a1c74d04f735d164fdc621e3ce3d5bfea1026243 | 223 | true |
INT008796 | S00072 | U00091 | 2025-03-18T21:39:16Z | train | internal_tool | customer_support | BR | AMER | free | team | support_strict | anthropic | claude-3.5-sonnet | 0.485 | 0.929 | 1,024 | 1,329 | 685 | 2,014 | 1,151 | 0.64448 | 0.804 | false | none | false | false | false | false | false | false | 0 | none | 1 | positive | false | 0 | none | auto_resolved | low | Quick question: I need a short, friendly reply to a customer who is asking about the status of their refund. Please explain step by step. | Here's what I'd implement in production. Goal: handle unexpected charge reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Prefer simple... | 1,749.782798 | 2025-03-18 | 21 | 1 | false | false | 1.9401 | I need a short, friendly reply to a customer who is asking about the status of their refund. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 240 | true |
INT008797 | S00838 | U00363 | 2025-03-25T10:43:52Z | train | web_app | customer_support | FR | EMEA | pro | individual | internal_policy_qa | meta | llama-3.1-70b | 0.56 | 0.922 | 1,536 | 533 | 800 | 1,333 | 659 | 0.2666 | 0.791 | false | none | false | false | false | false | false | false | 0 | none | null | no_feedback | false | 0 | none | auto_resolved | low | I'm stuck on this and need some guidance: A user says they were double charged this month. Help me draft a clear explanation and next steps. Please explain step by step. | Below is a practical plan you can execute. Goal: handle missing feature reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Start... | 2,022.76176 | 2025-03-25 | 10 | 1 | false | true | 0.6662 | A user says they were double charged this month. Help me draft a clear explanation and next steps. | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | false | 03bf57d7adc83026d525c8192379aa203b71e1eb | 236 | true |
INT008798 | S00019 | U00354 | 2025-04-03T15:39:38Z | train | mobile_app | content_writing | IN | APAC | enterprise | individual | code_assistant_secure | meta | llama-3.1-8b | 0.852 | 0.82 | 1,536 | 460 | 957 | 1,417 | 1,580 | 0.08504 | 0.924 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | auto_resolved | low | I'm stuck on this and need some guidance: Write a concise product description for a new wireless mouse aimed at software developers. A short, clear answer would be ideal. | Here are the guardrails I'd put in place. Goal: handle tool calling reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Start by writing a small... | 896.835443 | 2025-04-03 | 15 | 3 | false | true | 0.4807 | Write a concise product description for a new wireless mouse aimed at software developers. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | 20f1dad1d39f7b8df6faaa166fe78024f8d03fd6 | 251 | true |
INT008799 | S01552 | U00261 | 2025-03-19T19:56:23Z | train | web_app | content_writing | BR | AMER | free | team | general_assistant | meta | llama-3.1-8b | 0.563 | 0.874 | 256 | 1,006 | 136 | 1,142 | 1,800 | 0.06852 | 0.892 | false | none | false | false | false | false | false | false | 0 | none | 2 | positive | false | 0 | none | abandoned | medium | Quick question: Rewrite this paragraph in a clearer and more professional tone. I'd like a practical, engineering-focused answer. | Here's what I'd implement in production. Goal: handle LLM reliability reliably, with guardrails for formatting and failure handling. Plan: - Add telemetry fields for latency/tokens/cost so you can debug regressions quickly. - Prefer simple... | 634.444444 | 2025-03-19 | 19 | 2 | false | false | 7.3971 | Rewrite this paragraph in a clearer and more professional tone. | You are a writing assistant helping users craft clear, engaging, and well-structured content. Keep the tone natural, readable, and aligned with the user's intent. | true | a1c74d04f735d164fdc621e3ce3d5bfea1026243 | 242 | true |
INT008800 | S00549 | U00062 | 2025-02-22T15:30:07Z | val | api | customer_support | US | AMER | free | individual | support_strict | meta | llama-3.1-70b | 0.633 | 0.913 | 768 | 532 | 244 | 776 | 1,039 | 0.1552 | 0.62 | false | none | false | false | false | false | false | false | 2 | code_run|knowledge_lookup | 2 | positive | false | 0 | none | auto_resolved | low | Can you help me debug this? Our customer is getting an error when trying to log in to their account. Can you help me debug this? Please explain step by step. | Below is a practical plan you can execute. Goal: handle missing feature reliably, with guardrails for formatting and failure handling. Plan: - Use schema validation (e.g., Pydantic) to prevent bad inputs from propagating. - Start by writing a... | 746.871992 | 2025-02-22 | 15 | 5 | true | true | 2.1803 | Our customer is getting an error when trying to log in to their account. Can you help me debug this? | You are a helpful customer support assistant for a SaaS product. Answer clearly, empathetically, and step-by-step while keeping the response concise and practical. | true | d029d0919f8e306d09ca7001d500c1e6f1c81007 | 245 | true |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.