Turn the approved practices behind your team's repeated work into rules that guide each AI session, catch predictable mistakes, and recommend the right next step.
One example of Rulebook in action. The same pattern applies anywhere your team has an approved way to work.
Every repeated workflow has context the prompt leaves out: which environment to use, how to handle customer data, which implementation pattern is approved, and what to verify before acting. Those practices often live across people, chats, and runbooks instead of inside the active AI session.
The request captures the outcome, but not every team-specific boundary.
The approved practice is scattered across people, tools, and old documentation.
An action can look reasonable to the agent and still violate the way the team works.
For a customer demo reset, Rulebook can surface the production-safety practice and point the agent toward the approved alternative.
The demo reset is one example. Rulebook applies the same pattern wherever a repeated workflow has an approved way to work.
Illustrative product view · not a customer environment
The specific rule changes with the workflow. The pattern stays the same: capture the approved practice, make it available when the work calls for it, and help the agent follow the right path.
Preserve the approved steps for preparing, resetting, and seeding customer environments.
Surface approved configuration, data-mapping, authentication, and customer-data handling patterns.
Bring the right preconditions, environment rules, sequencing, and verification steps into the active session.
Turn corrections, incident lessons, and failed paths into guidance the next person or agent can use.
Add the rule, why it exists, and where it applies.
Make the guidance available when an agent reaches the relevant part of the workflow.
Point the agent toward the approved way to complete the task.
No. Static files stay tied to a particular repository, tool, or person. XTrace provides a shared Rulebook that can make relevant guidance available across connected agent workflows.
No. Rulebook complements permissions, testing, and code review. It does not replace them. Keep the controls your team already relies on — Rulebook helps the agent reach the work with the right approved practice, before downstream checks become the first place a mistake is discovered.
No. Rulebook is designed to make the same current approved practice available across the agents and interfaces your team already uses.
Choose one repeated workflow where the team already agrees on the right way to work. Add that practice to a Rulebook and test it in a real AI session.
Add that rule to XTrace, test it in a real AI session, and see whether it appears before a predictable mistake becomes rework.