News and updates
Product releases and company news.
PII detection now validates structured identifiers, such as account and card numbers, before treating them as personal data, and recognises a much wider set of credential and secret formats.
The built-in SOC 2 and NIST AI RMF rules now check for detected PII in any agent event rather than in model input only, and existing organisations can upgrade their framework rules in place. PII is detected and flagged; redaction remains on the roadmap.
The dashboard adds an investigation workflow for traces: start from a flagged or blocked action and see what the agent did, which rule matched, and what was decided, in one place.
Run gentrail with no arguments to open a guided setup menu: request or save a licence, deploy to your AWS account, connect an agent, and check the deployment.
After you connect an agent, the CLI waits for its first activity and opens it in the dashboard, so you know the connection works.
We are working with a small number of organisations to test runtime controls, assess compliance gaps, and prepare audit evidence in their own environments.
Each engagement starts with an agreed scope and success criteria. It is separate from the 60-day free evaluation, which you can start on your own at any time.
Replay a new or changed rule over the past seven days of traces to see what it would have caught before you turn it on.
Connect agents built with the OpenAI Agents SDK, LangChain, LangGraph, the Vercel AI SDK, and Strands, or observe any agent through OpenTelemetry.