Skip to content
Takeida

What you keep here can run things outside it

A Concept is reviewed, versioned and searchable, and it can be read by more than the people looking at it in the app. Neither of these is a separate feature. They are what the features you have already read about add up to, once something outside reads a Concept.

  • Skills and memory
  • A content backend
  • Scribe

    on your laptop

    Reads its skills here.

  • Reviewer

    on a server

    Keeps its memory here.

  • Your website

    Reader role

    Reads what is published.

Skills and memory

Give your agents a shared, governed memory.

An agent's skills and instructions usually live on the machine it runs on, one copy per machine. Keep them in a Concept and every agent you connect reads the same source. Change a skill once, from your phone if you like, and each agent has the current version the next time it reads, wherever it runs, with nothing to update locally.

Memory works the same way. An agent keeps what it has learned in a Concept, so it carries across sessions and across machines, and outlives the container the agent happened to be running in.

And because it is a Concept, it is governed. An agent proposes an addition; a person approves what becomes canon. You can read what your agents believe, see what changed and why, and correct it.

A content backend

Run a site, a list or a CRM from a Concept.

Use a Concept as the source behind a website, a customer list, a subscriber list or a CRM page. The people who write the content work in the editor, with drafts and review, and what is published is what the site shows.

The site reads through the Reader role. Give its app or its agent a key as a member with that role, on a Concept that is public or private, and it can read what is published and nothing else. It is a member in its own right, not a loan of somebody else's access, so it cannot write, propose or publish, whatever happens to it.