TraceX

What TraceX is

TraceX is an AI engineer for production issues. It reads the support email, Slack message or Jira ticket, investigates your database, code and logs read-only, finds the root cause with the evidence for each claim, and writes the fix as a pull request along with a reply for the customer.

How an investigation works

• Intake: an issue arrives from a Gmail support inbox, Slack, Jira or the dashboard.
• Evidence: read-only queries against PostgreSQL, MySQL or MongoDB, code and recent commits through the GitHub App, and Sentry, Datadog and CloudWatch logs.
• Root cause: a confidence score, expected versus actual behaviour, and citations to the records, code and log lines involved.
• Action: a small, reviewed diff opened as a pull request with validation steps, plus a drafted customer reply.

More than investigations

• Conversations: ask questions about production data and code in plain English, with exports to spreadsheet, CSV or PDF.
• Fixes and code review: fixes as pull requests and reviews on your team's PRs.
• TraceX CLI: one-line install on macOS and Linux (Windows builds available); answers in the terminal and a coding mode that asks before every edit and command.

Security

The company says TraceX makes no production writes, uses read-only credentials and refuses write or schema queries, keeps no index of your code or data, deletes exports after 24 hours, checks each answer with a second model, and never trains on customer data. SSO, 2FA, roles and an audit log are included. Teams can bring their own Anthropic or Gemini key.

Pricing

Free during the private beta: 50 investigations and 500 chat messages a month, up to 5 seats, all integrations and the CLI. Larger teams can arrange higher limits, private tenancy or self-hosting.