book a free evals working session with us

Get everyone working
from the same evidence

Connect what your agent did with the people who know whether it was right. neatlogs keeps production evidence, business judgment, team decisions, and engineering action together, from the first report to the verified fix.

Works where your team already does:

SlackDiscordIntercomGmailLinearJira
21st Aug, 12:57 PM
Search
support automation demo14 steps · 18s
14 steps · 18sExpand all
+northline_support_runWorkflowError
+triage_agentAgent
+lookup_customerTool
+gpt-5.5-2026-04-24Model
+knowledge_agentAgent
+search_kbTool
+render_policy_briefTool1
+resolution_agentAgentError
+get_orderTool
+case_media_boardTool2
+create_refundToolError4
Ended with error

PaymentsAPI 503: refund service unavailable

One span, one shared decision

Context becomes evidence

Connect business contextwith production data.

A trace can show that the agent issued a refund. It can’t tell you whether that customer was entitled to one. neatlogs connects production runs with the feedback, policies, exceptions, and decisions that define what correct meant in that case.

n
Context linkNorthline refund report
Live
Customer conversationIntercom · NorthlineNow
Sarah

SarahThe agent said my refund was approved, but this order was outside the return window. Which answer is correct?

Reported in the customer’s own words
Matched execution98% match
run_184221 Aug · 12:57 PM
+northline_support_runWorkflow
+triage_agentAgent
+lookup_customerTool
+knowledge_agentAgent
+render_policy_briefTool
+resolution_agentAgent
+create_refundToolMessage linked
Meaning preservedAttached
Original report

“The agent said my refund was approved, but this order was outside the return window.”

PolicyRefunds require eligibility first
Customer exceptionNo approved exception found
Business judgmentIncorrect promise, not merely a failed tool call

Comment on the exact thingthat’s wrong.

Highlight the evidence. Leave a comment. Tag the person who knows. Keep the answer where the question began.

01
nProduction trace14 steps
northline_support_run WORKFLOWresolution_agent AGENTcreate_refund TOOL1
Sarah

Sarah This promise is made before eligibility is checked.

On a span

Point to the prompt, tool argument, retrieval, or line of output that needs attention.

Comment stays attached
02
PII in outputsDetection
create_refundPossible account number

“…sent to account ending in 4821.”

Chris

Chris Useful signal, but mask the last four digits from this rule.

On a detection

Decide whether the signal is useful, noisy, or scoped incorrectly.

Comment stays attached
03
Refund policy investigationDraft
ROOT CAUSEThe eligibility guard was skipped before the refund tool ran.
RECOMMENDED ACTION

Require a policy outcome before issuing the refund.

Jon

Jon Add the enterprise exception before this ships.

On an investigation

Challenge the root cause, supply missing context, or review the recommended action.

Comment stays attached
04
Refund qualityEvaluation
Would we send this response to an actual customer?
AINoPolicy was not checkedHYesReviewer disagreed
Matthew

Matthew The rubric should require eligibility before any promise.

On an eval

Discuss the verdict, the rubric, or why two reviewers disagreed.

Comment stays attached

Use neatlogs without
opening neatlogs.

Review an eval, answer an investigation question, acknowledge an alert, or reply to a thread from Slack. The question arrives with the output and evidence needed to answer it.

Reply in Slack. The same thread updates in neatlogs.
Answer in Slack
Thread# agent-support
n
neatlogsAPP10:42 AM

The create_refund span failed with the following error:

PolicyViolation: refund attempted before return eligibility and support approval were confirmed.

Does the enterprise exception apply to this customer?

PRODUCTION EVIDENCErun_1842 · create_refund

Policy check did not run before the tool call.

2 replies
Matthew
Matthew10:45 AM

No. The exception requires support approval, and this run did not have it.

Chris
Chris10:47 AM

Agreed. I’ll add the approval check before the refund tool can run.

Evidence arrives in neatlogs
Issues ›Refund issued before eligibility was confirmed
Open
How often1 failing trace
4 SepToday
1 northline_support_run trace failed at issue_refund4 Sep
northline_support_runrefund_agentissue_refund PolicyViolation
Matthew replied in #support-agent.M1 message · 1 cited
Matthew

MatthewNo. The exception requires support approval, and this run did not have it.

Issue created — issue_refund — Refund RF-88244 for $240 issued on NL-10517 with return_eligible=false and vip=false

From feedback to a verified fix

01
neatlogs# agent-support
# agent-support4 members
Sarah
Sarah

The customer was promised a refund before eligibility was checked. Can someone look at this?

neatlogs
neatlogsAPP

I found the matching production run and attached the reported outcome.

Someone notices

A customer, support lead, PM, or engineer says something went wrong, in the tool they’re already working in.

02
nIssue raised
High severity
ISSUERefund issued before eligibilitySlack report · run_1842 attached
Slack messageProduction runIssue
Source preservedSeverity, reporter, and exact span attached

neatlogs finds the run

It connects the message to the runs it’s about, and raises it as an issue with severity and source attached.

03
Refund policy investigation3 comments
Root causeEscalation check was skipped

The refund path continued after the policy lookup returned an exception.

Sarah
Sarah

@neatlogs this account is in the enterprise exception group. Did the run load that policy?

Neatlogs
neatlogsAPP

Good catch. I found a second policy lookup and added it to the evidence.

The investigation uses their context

The reported outcome, business rules, and team decisions become part of the diagnosis.

04
LLinear issueCreated
SUP-482High severity

Require escalation before enterprise refunds

Prevent `create_refund` when the account policy requires a specialist review.

Evidence
4 spans + report
Acceptance
3 verification cases
Investigation linked to the ticket

The fix moves forward

The report goes to the owner as a ticket, or to your coding agent with the evidence intact.

05
Production watchLive
Fix deployed

Enterprise refund guard

Watching the same failure path in production.

Policy loaded248 runs
Escalation respected100%
No recurrence7 days
neatlogs will re-surface the issue if it returns

The loop closes

The person who raised the problem can see what changed. neatlogs watches your agent to check whether the fix held.

See how investigations work

People and agents
work the same case.

Every contribution stays on the same issue, with its author, evidence, and decisions visible.

ONE SHARED ISSUERefund issued before eligibility
nSarahClaude Coden
01n
Investigation agentneatlogs

neatlogs’ investigation agent assembles the evidence.

02Sarah
Domain expertSarah

A domain expert resolves the question only they can answer.

03Claude Code
Coding agentClaude Code

Your coding agent acts on the fix neatlogs generates and you approve.

04n
Production monitorneatlogs

neatlogs watches production and notifies you if the same failure returns.

Watching
Evidence, authors, and decisions remain attached to the same issue.
Start with production history

Get your whole team looking at the same run

Bring in traces from your current platform. Then connect the places where your team and customers already report problems. Or instrument directly.

Product questions and answers

The practical details about connecting team feedback, discussing evidence, and keeping decisions attached to production work.

Selected question01 / 09

Do reviewers need to understand traces?

No. Someone answering a question in Slack sees the output being judged and the evidence behind it, not a span tree. The people who know what correct means at your company are usually not the people who read traces.

No. Someone answering a question in Slack sees the output being judged and the evidence behind it, not a span tree. The people who know what correct means at your company are usually not the people who read traces.