Ren 1.1, by invitation.

A decision model, served through an API. The private beta starts with a narrow contract and room for human judgment.

Give the model a decision

Most applications do not need another paragraph. They need to choose where a request goes, how a case should be handled, or whether a condition holds. Ren 1.1 takes the situation and the decision you define. It returns a structured answer with probabilities over the options you offered.

The private beta supports a choice between named options, an ordered score, and a true-or-false check. The model is served only through the API. Your application stays responsible for validation, permissions and what happens after a result arrives.

Keep people in the loop

Each request carries a gate threshold. The server tells you whether that threshold was met or whether the case should go to a human. It does not certify that the chosen action is right. Test thresholds on representative data and review the cases that matter.

Probabilities cover the available options. If every option is unsuitable, a probability map does not discover an extra escape route. Include an explicit escalation choice or use a companion true-or-false check when that possibility matters.

Start small

Use a pinned model release when you compare behavior. Keep inputs focused. Try the playground, then carry the same request into your own code. Review mistakes before connecting the result to an action.

Access is by invitation while we work with developers on their decision tasks. Customer inputs and outputs are not used to train our models. Contact us with the workflow you want to evaluate; the docs explain the request contract and its limits.

All notes →