- +northline_support_runWorkflow
- +triage_agentAgent
- +lookup_customerTool
- +gpt-5.5-2026-04-24LLM
- +knowledge_agentAgent
- +search_kbTool
- +resolution_agentAgent
- +create_refundTool
neatlogs investigates.
You decide what ships
neatlogs gathers the evidence, finds the root cause, and writes the fix brief. You review it before anything changes
Findings go where your team already works:
SlackLinearJiraGitHub
Claude Code
CursorMCPEvidence inSlack conversationCustomer impact and context
Production tracesRuns · spans · tool calls
Slack conversationCustomer impact and contextIssue contextLinear · Jira · support
Change historyDeploys · commits · config
Cause isolatedEvidence agrees
Fix brief ready forClaude CodeAct on the briefCursorPrepare the changeCause · evidence · next steps
Claude CodeAct on the briefCursorPrepare the changeGitHubReview the pull request
Every incident starts with
the evidence attached
A detection fired
A rule, threshold, regex, or classifier flagged the exact span where the behavior appeared.
An eval failed
A human or AI evaluator judged the output against your standard. And it fell short.
neatlogs
@neatlogs Refunds are failing again for Northline. Can you check the latest run?
Matched production evidencesupport_agent.run
run_284 · customer_history · ignoredSomeone reported it
A customer or teammate reported a bad outcome in Slack, Discord, email, or Intercom.
Failed decision
Repeat-delivery exception was missed
Two prior delivery failures were loaded but not used.
Customer history loaded2 prior delivery failures
Three retriesExponential backoff
Execution failedService unavailable
neatlogs found it
neatlogs spotted a broken execution or suspicious behavior without waiting for a predefined rule or report.
Then neatlogs
works the case
neatlogs investigates everything already known about the incident.
northline_support_runrun_1842AIrefund_agentissue_refundPolicyViolation
northline_support_runrun_1848AIrefund_agentissue_refundPolicyViolation
EvaluationRefund policy review
100%4 items
One timeAI
Slack · #support-opsNow
Sarah
This customer has two prior delivery failures. The agent refused the current refund without using that history.
Chris
customer_history completed in both runs, but decide_resolution never received the result.
One report,four answers
Investigation reportSupport agent · 05 Sep
92%
High confidenceEvidence and context agree
Retry removedcreate_checkoutv2.8
See what broke
Follow the evidence
Know what to change
Verify the fix holds
Review it. Ship it.neatlogs sees if it held
Root causeEscalation check was skipped
The refund path continued after the policy lookup returned an exception.
This is wrong. The customer had two prior delivery failures, so the current undelivered order should be refunded. @Mark can you confirm?
Confirmed. After two delivery failures, the repeat-delivery exception applies to the current order. @neatlogs
Good catch. I added that context to the evidence and updated the investigation.
SarahCommenting on the root causeMarkReplying to SarahStep 01
Discuss the investigation
Comment on it, question the reasoning, and bring in the people with missing context.
Investigation V1
77%
Medium confidence
Root cause
Issue tracker
Linear
Jira
LinearJiraSUP-482
Require escalation before enterprise refunds
Investigation report4 evidence spans attachedLinked
Step 02
Create the ticket
Send the incident to Linear or Jira with its severity, evidence, fix, and acceptance criteria.
77%
Medium confidence
Root cause
Ready for implementation4 files
Open in
Cursor
Codex
Claude Code
GitHub Copilot
CursorCodexClaude CodeGitHub CopilotStep 03
Hand it to your coding agent
Send the fix to Claude Code, Cursor, or another MCP-compatible agent.
Investigation V1
77%
Medium confidence
Root cause
New context in #neatlogs-engineeringThe repeat-delivery exception applies after two failed deliveries.
NewStep 04
Reopen the case
Run the investigation again when a teammate adds context, or new evidence changes the conclusion.
Total traces248↑ 18.4%
Error rate0.0%↓ 100%
Detection rate100%↑ 24.6%
No failures after deploy
Fix is holding in production
Step 05
Verify it in production
neatlogs watches the behavior after the fix ships and resurfaces the issue if it returns.
One system.Four different questions
CapabilityThe question it answers
01Trace
What happened in this run?
02Detection
Where did a known condition appear?
03Eval
Did the behavior meet an explicit standard?
04Investigation
Why did it happen, and what should change?
FAQs01 / 11
What’s an incident?
An incident is a flagged problem in an agent’s behavior. It carries a severity, a source, and the runs it belongs to. An incident can come from a fired detection, a failed eval, a person reporting something, or neatlogs raising one on its own.
FAQs
An incident is a flagged problem in an agent’s behavior. It carries a severity, a source, and the runs it belongs to. An incident can come from a fired detection, a failed eval, a person reporting something, or neatlogs raising one on its own.
Still unsure?
We’ll look at your stack with you.