How do you turn scattered team knowledge into skills AI agents can actually follow?

by•

Software teams have a lot of valuable knowledge spread across Slack, Teams, Jira, Notion, Google Drive, documentation and people's heads.

AI tools are getting increasingly good at searching these sources and answering questions. But there's another problem I'm interested in.

How do we turn what a team knows into something AI agents can consistently follow and execute?

I'm building Lynxient, an intelligence layer for software teams, around this idea.

The workflow we're working on connects four things:

  • Knowledge retrieval: Ask questions across connected tools and get answers linked to their original sources.

  • Process extraction: Turn recurring conversations and documents into SOPs that a designated owner can review and verify.

  • Agent skills: Compile verified SOPs into reusable skills that Claude Code, Cursor and other AI agents can follow through MCP.

  • Controlled automation: Let agents prepare and execute workflows, with human approval before external actions are committed.

There are already some great products tackling different parts of this space. Glean, Guru, Dust and Unblocked are all interesting examples. Each approaches company knowledge, AI agents and context from a different angle.

What I'm trying to understand is where the gaps are between finding company knowledge and actually making that knowledge useful to autonomous agents.

A few things I'd love to hear your thoughts on:

  1. How are you currently giving AI coding agents access to your team's decisions, internal processes and operating procedures?

  2. Would you find it useful to have verified SOPs automatically converted into skills for Claude Code or Cursor?

  3. What is the bigger challenge today: getting agents the right context, keeping that context accurate, or making sure they execute the right actions?

  4. How much human approval should be required before an AI agent takes action across your company's tools?

I'm particularly interested in hearing from founders, CTOs, engineering leads and people already using AI agents in their daily workflows.

Curious to hear what you're using today, what's working, and where things are still breaking.

8 views

Add a comment

Replies

Be the first to reply

Have a question or a thought to share? Add a comment above to start the conversation.