For teams: coding agents on the self-hosted models you're allowed to run - see early access →
polyglot

Early access

Coding agents on the models you're allowed to run

Your developers want AI coding agents. Your code can't go to a cloud LLM. And the open-weight models you can self-host often skip or garble the tool calls an agent depends on. We're building a self-hosted layer for teams in that position: qualify each model against your own tools, run agents reliably on the ones you approve, and keep an audit trail of what they did.

No account created, nothing charged - just lets us reach out as this takes shape. Your email is only used for that; it's never shared or used for anything else. See the privacy page for details.

What it includes

Some of this works today; the rest is what we're building next. Each item says which.

Qualify

Know which of your models can be trusted with agent work before you approve them.

  • Test a model or quantization against your own tool schemas, in your environment, and get a signed reportIn development
  • Compare a newer model against the one you've approved, so upgrades don't stall in reviewIn development
  • Flag a model for re-qualification when its weights, quantization or behaviour changeIn development

Run

Coding agents that keep working on the open-weight models you're allowed to run.

  • One shared OpenAI-compatible endpoint your team's agents point atWorking today
  • Recovers the malformed or missing tool calls that make agents stall on self-hosted modelsWorking today
  • A reliability scorecard: which models complete tool calls, and how often they need repairWorking today
  • Drift alerts and a check for misconfigured chat templates on your serving stackIn development

Govern

See what agents did, and which calls ran on a repaired rather than clean model output.

  • Team-wide activity feed and audit trail: which tool calls happened, by whom, on which modelWorking today
  • An inventory of the agents, tools and models in use, discovered from trafficIn development
  • Policy as code for what agents may do, run in monitor mode before anything is blockedIn development

Built to pass a security review

It works on its own - no other AI platform required. Planned for the team release:

Self-hosted, including air-gapped networks. Nothing about your code or models leaves your infrastructure, and there's no phone-home.
Sign-in through your identity provider. OIDC SSO on every plan; SAML and SCIM on Enterprise. No separate passwords to manage.
Policies are versioned files in your own repo, signed with your own key. We hold no key that can change your policies.
Exports to the tools you already run: JSON Lines, syslog and OpenTelemetry.

Indicative pricing

We haven't finalized this - showing a real number here so early signups reflect actual interest, not just curiosity. Tell us if it's off.

Starter

Up to 15 developers

$29/developer/mo

  • Shared endpoint with tool-call recovery
  • Reliability scorecard and drift alerts
  • Qualification report for one model at a time
  • OIDC single sign-on

Team

16-75 developers

Talk to us

  • Everything in Starter
  • Activity feed and audit trail
  • Model approval workflow and re-qualification
  • Policy in monitor mode, per-user budgets, SIEM exports

Enterprise

75+ developers, or any regulated team

Talk to us

  • Everything in Team
  • Policy enforcement and approvals
  • Signed evidence exports
  • SAML, SCIM and an air-gapped install

Get early access

We'll reach out as this takes shape - to ask what you'd need, not to sell you anything yet.

Your email is only used to reach out about this - never shared or used for anything else. See the privacy page for details.