Check the work, not just the answer
See how a result was reached: the steps, the sources and the files behind it.
Run trace and audit
See exactly what your agents did, and why.
Every run shows its steps, sources, files, retries and changes, in a record your team can read.
Workspace
Run trace
Request received Complete Policy matched Complete Usage summary checked Verified Legal clause compared Verified Approval requested Waiting
A record your whole team can read.
Find the problem fast
When something fails, the trace shows where it stopped, what was retried and what blocked it.
Hand auditors a clear record
Every action and approval is kept in one readable record, ready to review or export.
A step by step view of every run
Each run is laid out in order: how the request was classified, which scope was checked, what evidence was gathered and where it waited for approval.
- Every step with its time and result
- Retries, blockers and failures in place
- Approval points marked in the run
- 09:41 Request classified
- 09:43 Scope checked
- 09:48 Evidence gathered
- 10:02 Approval waiting
Artifacts Contract delta.pdf Spend summary.csv Recommendation.md
Evidence attached to each step
Sources, files and artifacts stay attached to the step that used them, so a reviewer can open exactly what the agent saw before it acted.
Workspace
Northstar BI Renewal Owner Brief
Marcus Lee
Northstar BI sent a 12% renewal uplift. Keep Finance, Procurement, and Legal in one thread.
Renewal Research Agent
Owner brief is ready. Approval is required before vendor outreach.
Medium risk 7 events logged
Audit log and output history
Actions, tool calls, knowledge checks and approvals are written to an audit log. Earlier outputs stay in history, so you can compare versions and see what changed.
- Role based access to records
- Output history for every run
- Records you can review and export
Controls RBACSkill accessModel policyApproval rules
Audit log
User approved packet
Agent called CRM tool
Knowledge source checked
How it works
From request to recorded result.
-
A request arrives
From the inbox, a thread or a scheduled task.
-
The agent plans
It sets out the steps it will take and the tools it needs.
-
Each step is logged
With its sources, files, retries and results.
-
People review and approve
Sensitive steps pause, and every decision joins the record.
-
The full record is saved
The finished run keeps its complete trace for later review.
Every feature runs inside the same six controls.
Identity, access, approvals, supervision, audit and policy apply to every agent and every run, whichever part of the workspace it happens in.
See governance and audit- Identity
- Access
- Approvals
- Supervision
- Audit
- Policy
FAQ
Run trace and audit: questions and answers
Short answers to what teams ask before they put this part of SoftworkerAI to work.
Contact us directlyWhat is a run trace?
A run trace is the step by step record of one agent run. It shows what the agent was asked to do, the plan it made, the sources and tools it used, where it paused for approval and what it changed.
How long are records kept?
Run records stay in the audit log of your workspace, next to the work they describe. During early access we agree retention with each team as part of setup, so it matches your own record keeping rules.
Can we export the records?
Yes. Every run is logged with evidence you can review and export, so security, compliance and audit teams can work with the record outside the workspace.
Who can see a run trace?
Access follows the roles and permissions in your workspace. People see the traces for work they are allowed to see, and admins can review activity across the workspace.
Does the trace show failed runs?
Yes. Failed and blocked runs keep their trace, including where they stopped, what was retried and what error or blocker they hit, so the problem is quick to find.
The multiplayer workspace for humans and AI agents is how teams move beyond prompting and safely hand AI real work. It gives people one place to plan, approve, and oversee execution, so they can run work with AI agents they can govern and trust.