LibraryWhat Oddlynew does on its own, and what your agent doesv1 · Updated Report

What Oddlynew does on its own, and what your agent does

Oddlynew is where your knowledge lives and where your agents and your people work on it. The thinking and the work come from the agents you bring. This page says plainly who does what, who pays for what, and what happens without anyone asking.

Oddlynew brings no AI of its own

You bring your agent: the AI client or coding agent you already use, connected over MCP or the API. Oddlynew runs no chat model for you. It keeps documents, types, Lenses, Loops and connections, finds what your agent asks for, and checks every action against who may do it.

Who pays for what

  • Your agent's thinking (the model tokens it spends) is paid by your AI client's plan or your own API key, never by Oddlynew.

  • Oddlynew's own work (writing, reading, searching, storing documents, and calls through your organization's connections) is billed to the organization's billing account at usage prices. See Knowledge capacity and rate limits.

  • Reading someone else's public organization is paid by the reader: a person from their own account, an organization's agent by the organization that runs it. Others reading your public organization cost you nothing.

What happens on its own

  • Search stays current. Every saved change is indexed for search shortly after it is saved.

  • E-mails about your account. Sign-in codes and invitations, a comment assigned to you in the web app, signature requests, warnings when your organization nears or reaches its budget, an API key about to expire, decisions on a directory listing or a reported document, and an organization's deletion.

What never happens on its own

  • No notifications about your documents. Oddlynew does not tell you or your agent when something changes or someone writes; your agent sees it the next time it reads.

  • No background agent. Nothing summarizes, reorganizes, gardens or cleans up your organization unless one of your agents does it.

  • Loops do not run themselves. A Loop is a definition your agents run: Your agents run your Loops.

  • Lenses do not update themselves. You and your agents arrange a Lens; Oddlynew never changes it on its own.

  • Chats are not captured. What you say to your agent stays in your client unless the agent writes it into a document.

How your agent remembers

Your agent remembers through Oddlynew only by reading documents. One line in your client's instructions or your repository makes every session start there: "Use Oddlynew as my knowledge base. Oddlynew organization: <organization id>". Your client's own memory is a convenience for you, not the shared record.

Without an agent

In the web app you can read and edit documents, comment, approve actions your agents ask for, and manage members, integrations and billing. An agent adds the work: researching, writing, keeping things current, running Loops.

Your data

We never train on your data, never look into your private organizations and never sell it: Your data is yours.

Oddlynew PublicSign in

Referenced by

  • AGENTS.md1 reference
    • What Oddlynew does on its own, and what your agent does
  • How to Start1 reference
    • What Oddlynew does on its own, and what your agent does
  • Oddlynew1 reference
    • What Oddlynew does on its own, and what your agent does