Trust, privacy, and compliance

Reading and exporting the per-decision audit trail

Every decision the Unless support agent makes leaves a permanent record. This record is what lets you prove what the agent did, why it did it, and which sources it relied on. Whether you are responding to a customer dispute, preparing for a regulator, or just reviewing a tricky conversation, the audit trail is the single place to look. The system logs inputs, retrieved sources, the model used, output text, and any procedures triggered, all tied to a timestamp and user identity where applicable.

Where to find the audit trail

The per-decision audit trail lives inside the Conversations view. Open any conversation and you will see a decision trail panel that shows the full record for that interaction. This is the same view your DPO or internal auditor can access if you give them the right role.

For a workspace-wide view, you can also use the audit export in Trust. This pulls the broader system audit trail, which records configuration changes, integration changes, and access events. Each entry shows the actor, the change, and the timestamp.

Exporting the audit trail

To export a single conversation, open Conversations, find the conversation you need, and click “Export audit trail”. You can download it as PDF or JSON. The JSON format is useful if you want to process the data programmatically or feed it into another system.

For a full export of all audit data, use the audit export in Trust. This is the option to choose when an inspector or regulator asks for a workspace-wide record. Logs are retention-managed under your privacy settings, with standard retention for security-relevant logs set to one year unless your contract specifies a shorter period.

Giving your DPO or auditor access

You do not need to hand out full admin rights to share audit data. Create a DPO role in Team and roles with read access to Trust and Conversations only. The same approach works for external auditors: create an external-auditor role with the same restricted read access. This keeps the audit trail visible to the people who need it without exposing broader platform controls.

What is logged for each decision

Each decision in the audit trail includes the following:

Inputs, including the customer message and Living Context state
Sources retrieved
Model used
Output text
Procedures triggered
Timestamp and user identity, where applicable

This level of detail is what makes the trail useful for compliance. The EU AI Act assumes you can show why the agent did what it did, and the per-decision log is the mechanism that makes that possible. Unless is the Provider of the AI system under the Act, and the platform is configurable for AI Act obligations, including transparency notices and audit trails per decision.

Using the audit trail for compliance

When you need to prove EU AI Act compliance for a specific Moment, combine three things. First, the per-decision audit trail from Conversations. Second, the configuration status in Accountability, which shows which AI Act controls are active. Third, the Provider documentation in the Compliance Center, including risk classification and Provider obligations. Together these cover what an auditor will ask for.

The audit trail also supports incident response. Every meaningful action in the platform writes to the audit log, including conversation decisions, configuration changes, integration changes, and access events. Logs are protected against tampering, monitored continuously, and queryable for investigations or audits.

Conclusion

The per-decision audit trail is your primary tool for transparency and accountability. Whether you are exporting a single conversation for a customer dispute or pulling a full workspace export for a regulator, the process is straightforward. Set up the right read-only roles for your DPO and auditor, and you have a complete, exportable record of every decision the agent has made.