Agent Serve · Run history
See what each AI agent did, in order: the input, every block, the model and tokens it used, the cost and the output or error.
Agent Serve Run history is the logs module of Agent Serve, the AI workspace where teams build, deploy, and manage AI agents. It lists every workflow and agent run with its trigger, status, duration and cost, and opens each one as a block-by-block trace with inputs, outputs, tokens, errors and a snapshot of the workflow as it ran.
How it works
Agent Serve records every run as it happens, so finding a problem starts with a filter, not a search through inboxes.
Filter by time, status, workflow, folder or trigger, or search with level:error and duration:>5s.
Each block shows its input, output, model, tokens, retries and how long it took.
View Snapshot opens a frozen copy of the workflow, even if it has changed since.
Troubleshoot a failed run in Chat, then retry it from the run’s menu.
In depth
Agent Serve keeps the detail of each run in one panel, so a builder, an auditor and a support lead all read the same record.
View Snapshot in Agent Serve opens the workflow exactly as it was when the run started, so a later edit never hides what happened.
A failed run in Agent Serve offers Troubleshoot in Chat, which opens Chat with the run already in view.
When an Agent Serve table fills a column with an agent, the run behind each cell opens in the same details panel.
Agent Serve log destinations copy workflow logs to your storage or monitoring tools every hour or every day.
The list
Agent Serve Run history understands level:, cost:, date:, duration:, trigger:, workflow: and folder: in one search box.
cost:>2 date:this-week
Agent Serve logs runs started manually, by API, schedule, Chat, webhook, MCP, another workflow or an integration such as Slack.
All triggers → Schedule
The Agent Serve run list refreshes every ten seconds, and an open run keeps updating until it finishes.
Running → Completed
Switch Agent Serve Run history to Dashboard for runs, errors and latency over time, and success rates per workflow.
Runs · Errors · Latency
Agent Serve shows each run’s cost with its input and output tokens, and sorts the list by cost.
Sort by Cost
People with edit access export the filtered Agent Serve run list, traces included, as a CSV file.
logs-2026-10-07.csv
Examples
Three ways financial services teams read Agent Serve Run history.
A model risk reviewer filters Agent Serve Run history to one agent and a month, then exports the runs with their traces as CSV.
An operations lead searches level:error for last night’s scheduled runs, opens the failing block and retries once the source is back.
A claims manager opens the run behind a disputed summary in Agent Serve and reads which documents and prompts produced it.
Examples show what teams can build in Agent Serve. They describe patterns, not named customer deployments.
By the numbers
9
Built-in trigger types, plus each integration trigger
10 s
Refresh interval of the live run list
7
Log destinations
1 day–5 years
Log retention you can set, or forever
Figures are read from the product code.
Before and after
| Work | Before | With Agent Serve |
|---|---|---|
| When an agent gets it wrong | BeforeRerun it by hand and hope the problem shows up again. | With Agent ServeOpen the Agent Serve run and read the exact input and output of each block. |
| What the workflow looked like | BeforeNobody is sure which version ran last Tuesday. | With Agent ServeThe run’s snapshot shows the workflow as it was at that moment. |
| Spend on models | BeforeOne monthly bill with no link to the work behind it. | With Agent ServeCost and tokens on every run, sortable and exportable. |
| Audit requests | BeforeScreenshots pasted into a document. | With Agent ServeA CSV export, or logs sent to your own storage every day. |
Under the hood
| What is logged | Workflow runs and Agent Serve agent jobs, with trigger, status, duration and cost. |
|---|---|
| Trace detail | Per block: input, output, provider, model, tokens, cache use, retries, tool calls and errors. |
| Filters | Time range, status, workflow, folder and trigger, plus field:value search. |
| Views | Run history list and a Dashboard of runs, errors and latency. |
| Run actions | Retry, Cancel Run, Copy Run ID, Copy Link, Open Snapshot. |
| Export | CSV, for people with edit access. |
| Retention | Log retention from 1 day to 5 years, or forever, per organization or team space. |
| API | List runs and read a run’s detail with an Agent Serve API key. |
Governance
Agent Serve lets admins decide how long run logs live, what personal data they hold and where copies go.
Security and governanceAgent Serve log retention runs from one day to five years, or forever, set per organization or team space.
Agent Serve can redact personal data at the Logs stage, so names and account numbers never land in a trace.
Every workflow run records each block’s inputs, outputs and errors, so a reviewer can trace how a result was reached.
Proposals, approvals, declines and executed actions are written to the workspace audit log, which admins can export.
Admin, write and read roles per workspace, plus permission groups that limit which models, blocks and integrations a team can use.
Retention from one day to five years, deletion and access controls support GDPR obligations. Ready means the controls are in place, not a certification.
Works with
Every workflow run lands in Run history, with a snapshot of the workflow that ran.
Troubleshoot a failed run in Chat, and open run details inside it.
Open the run behind any cell an agent filled in a table.
Retention, PII redaction and log destinations apply to run logs.
Further reading
A regulator’s letter becomes a dozen questions, a dozen owners and one deadline. Here is a setup where a workflow logs and splits the request, a Relay drafts from approved sources and a named person signs off every answer.
Regulatory affairs · 13 min readEvery week the committee secretary rebuilds the same pack from memos, spreadsheets and emails. Here is a setup where four workflows assemble it from the source systems and the secretary decides when it goes out.
Commercial lending · 12 min readMost security questionnaires ask the same two hundred questions in a different order. Here is a workflow that drafts each answer from your own policies and past approved answers, and sends anything uncertain to the right expert.
Sales and security · 13 min readQuestions
Agent Serve records the trigger, status, duration and cost of each run, and a trace of every block with its input, output, model, tokens and any error.
Yes. View Snapshot in Agent Serve opens a read-only copy of the workflow as it was when that run started.
Not today. Agent Serve shows failed runs in the in-app notifications, and you can filter Run history by status. Log destinations can copy logs to your monitoring tools.
Yes. The Agent Serve API lists runs and returns a run’s detail with an API key, and log destinations send logs to storage or monitoring tools on a schedule.
Last updated
Enterprise access
Agent Serve is available to teams by invitation. Tell us what you want to build and we will set up your workspace with you.