Manny Silva is Head of AI Docs Practice at Promptless, where he helps teams optimize their documentation workflows. He has built documentation for Apple, Google, and startups of various sizes. He codified the Docs as Tests strategy and created Doc Detective, an open-source tool for testing documentation. He wrote Docs as Tests: A Strategy for Resilient Technical Documentation and its follow-up, Docs as Tests & AI: A Strategy for Self-Healing Technical Documentation. He likes diving into the deep end as the zeroth user.

From 2 AM Failure to 8 AM Fix: Self-Healing Docs, Agent Instructions Included

A product release renames a button at 11:47 PM. Nobody told you. Nightly documentation tests catch it at 2 AM. At 8 AM the fix is waiting for approval, diagnosis attached. Review takes 3 minutes. It used to take 45.

This talk walks a failure through all five steps of the self-healing loop, then adapts it to common issues: changed APIs, broken links, stale screenshots, and conceptual pages that mislead without being wrong. We’ll also climb a maturity ladder from “tests file a ticket” to “it merges its own trivial fixes” and discuss where to stop.

Another doc set belongs in that loop, and most teams don’t consider it yet: AGENTS.md, skills, and agent definitions. Those docs make testable claims and drift too, but when they rot, your agents don’t tell you. They confidently do the wrong thing, and tracing their errors can take days.

You’ll leave with a self-healing system you can build in a day, the metrics that say it’s working, and a list of what not to automate.

In this session, you’ll learn how to:

• Trace a documentation failure through detect, diagnose, repair, verify, and report, naming what each step produces.
• Adapt the same five-step loop to API changes, broken links, stale screenshots, and conceptual drift.
• Place their team on the five-level self-healing maturity model and identify the next level worth targeting.
• Assemble a working detect-and-diagnose pipeline in about a day from a testing tool, CI, an issue tracker, and an AI agent.
• Treat an agent instruction as a testable document: extract its assertions and choose the right grader for each.
• Govern an instruction library so it stops accumulating duplicate, untriggerable, and stale entries.
• Monitor accuracy, rollback, over-generalization, and fix rate, and demote a system’s autonomy when those numbers slip.