Skip to content

Concepts

Saylek connects the AI models running on machines people already own, so they can use them from their own applications. Personal use is free. You share your machine with people you choose, and people you know can share theirs with you.

Five ideas carry the whole product. Read this page before Quickstart if you want to know how it works; read Privacy and egress before you send anything sensitive.

Sharing is between two people

Sharing needs no group and no membership. It is an arrangement between you and one other person, and it points in one direction at a time.

  • You share with someone. They can send requests to a machine of yours. Nobody else can, unless you joined a group (one an organization runs or one anyone can join), whose members can too.
  • Someone shares with you. You can send requests to a machine of theirs.

Those are two separate decisions, and neither implies the other:

  • Inviting someone to use your machine does not share theirs with you. An invitation is something you give.
  • Asking someone to share grants access from them to you, not the other way round. You are asking to receive, and answering the ask is how they give it.

Sharing can run both ways between two people, and it does not have to. Each direction is a separate act by the person whose machine it is.

Machines and models

A machine you connect runs models. When you share it, the person you shared with can reach the models it runs, and nothing else. There is no shell, no filesystem, and no way to run arbitrary code on it.

Sharing a machine is always your explicit choice: saylek host wizard sets it up, and you can switch it on or off from the machine's page or the CLI. Using a model someone shared with you never turns your own machine into one that serves other people.

Work you send from your own machine to its local Saylek API takes priority there. If it arrives for a model that is serving someone you share with, their request is stopped before it finishes, rather than immediately.

Requests run on someone else's machine

This happens without any setting, so understand it before you send anything sensitive.

When you ask for a model your own machine cannot serve, the request goes through Saylek's registry to a machine that can: another machine of your own, one belonging to someone who shares with you, or, if you joined a group (one an organization runs or one anyone can join), one belonging to anyone in it. Nowhere else, and there is no toggle to find. To leave a group, run saylek circles leave <id> (find the id with saylek circles list), or turn on local-only mode.

That means your prompt, and everything your tool put in the context window with it, leaves your machine and is handled in the clear by the machine that serves it while it computes the answer. Receipts prove where a request ran. They do not hide its content from the machine that ran it.

One control bounds this. Local-only mode stops Saylek routing requests through your machine to any other machine, persistently or per request. It is read on your machine, so it does not cover a request an application sends straight to Saylek's hosted API. Having nobody share with you does not bound this either. Local-only governs Saylek's own routing; it does not override a proxy upstream you configured yourself. See Privacy and egress.

Receipts

A request that goes through your machine and completes leaves a signed record of the machine that ran it. When another machine ran it, that machine signs the record and your machine co-signs and keeps its own copy. A request that fails or is stopped part-way leaves no record, and neither does a request an application sends straight to Saylek's hosted API. The record is not there because anyone distrusts anyone. It lets you check where a request ran, and it is yours to replay and verify offline.

Fees

Saylek is free for personal use. Organizations can arrange a one-month free trial with us, and any paid service is agreed separately in writing. Sharing is not paid: sharing your machine earns no money and no credits, and access to someone else's machine comes only from their permission.

Next steps

Last checked 2026-09-24 · read as markdown at /docs/concepts.md