/connect

Grid in the harness.

Connect puts the factory in the coding agents and IDEs your engineers already use. Smart Router picks a capable lane from the Grid catalog, or you pin an id. Same OpenAI-compatible call. Invitation while we scale.

How it works

/how

One endpoint in the harness

Connect points the coding agent or IDE at Grid. Developers keep the tool they already use. You do not rebuild the harness. Setup is reversible.

Smart Router does the choosing

Set the model to grid-auto and Smart Router scores the request against the Grid catalog. Pin an id when you need a specific model. There is no surcharge for routing. You pay for the model that answered.

Spend you can see

Account default limit, per-user override, and usage by day, user, and model, in tokens. Finance can export CSV. We quote Connect as part of the invite. We do not publish a blended savings rate we cannot source.

Connect, Smart Router, Grid

/stack

Grid is the factory

Serve, route, fine-tune. Connect is how that factory shows up inside coding tools. It is not a second cloud and not a spend overlay that replaces Grid.

Smart Router is the engine

The same opt-in router as the API. Connect does not invent a second policy. If you already route with grid-auto in production, Connect is the same decision in the harness.

Not a default

Routing stays opt-in. A team can pin S Series or an open-weight id for the work that must not move, and route the rest.

Invitation

Connect is not public self-serve. Join the waitlist or talk to us. We will not list harness logos we have not shipped.

Your first call is one base_url away.

Request access

Grid is invite only while we scale.

Talk to us

Tell us who you are and what this is about.