AITracer
Docs
First Trace

First Trace

Capture your first trace and follow it through governance, verification, and review workflows.

Your first trace is where AITracer becomes operational.

After installation and integration, this walkthrough shows what happens when your application sends its first AI request through AITracer.

By the end of this guide, you will understand how a single execution moves through trace capture, governance review, verification, and long-term storage.

End-to-end trace workflow

Rendering diagram...

Step 1: Send your first request

Your application sends a model request through the AITracer SDK or Trace Ingestion API.

This can include:

  • prompts
  • model requests
  • tool calls
  • workflow metadata
  • user actions
  • downstream operations
from aitracer import AITracer
 
client = AITracer(
    api_key="your-api-key"
)
 
trace = client.trace(
    model="gpt-4o",
    input="Summarize this quarterly report",
    metadata={
        "workflow": "finance-review"
    }
)
 
print(trace.id)

The returned trace ID becomes the primary identifier for future investigation workflows.

Step 2: Review the trace

Open Trace Explorer to inspect your first execution record.

Review:

  • trace ID
  • prompt input
  • model response
  • execution timestamps
  • latency metrics
  • token usage
  • cost attribution

This confirms your trace was successfully captured.

Step 3: Governance review

The Governance Engine automatically evaluates the execution.

This may trigger:

  • PII detection
  • credential detection
  • policy validation
  • cost threshold monitoring
  • workflow restrictions

Low-risk executions continue normally.

High-risk executions can be flagged for review.

Step 4: Verify record integrity

The Verification Layer generates integrity metadata for the trace.

This includes:

  • SHA-256 fingerprint generation
  • timestamp validation
  • record verification
  • tamper detection checks

This ensures execution records remain trustworthy over time.

Step 5: Store the record

Verified records are stored in the Audit Vault for long-term retention.

These records support:

  • compliance audits
  • legal reviews
  • internal investigations
  • customer disputes
  • operational reporting

What happens next?

Once the first trace is working, teams typically expand AITracer across:

  • multi-agent workflows
  • internal copilots
  • customer-facing AI systems
  • automation pipelines
  • enterprise applications

A single trace becomes the foundation for governance, verification, and operational visibility.

InstallationTerminology