A single open server rack with glowing status lights against a concrete wall

    Runs where
    your software runs.

    On-premise, air-gapped, on your own model. Even nowhere near us.

    Talk to us

    Tell us about your setup.

    Targets, how many run at once, constraints. A person replies within one business day.

    • A demo, live on your screens
    • A pilot in your real environment and CI
    • Conversion terms in writing before it starts
    • Security questionnaire and DPA on request

    Used only to answer you. Privacy policy.

    Before you send it: where does your data go?

    Your data

    Only the inference payload leaves the machine. With a self-hosted model, not even that.

    • ISO 27001Certified
    • GDPRCompliant, DPA available
    • Your dataNever trains models
    • SecretsOS-encrypted, redacted
    • Air-gappedSupported
    • Your modelNo inference fee

    It stays with you. So where can it run?

    Deployment

    Four ways to deploy.
    One app.

    Nothing about writing or running a test changes between them.

    • HubSign in and go. Inference routed by AskUI.For teams starting out
    • Hub + your modelYour endpoint. No inference fee from us.For existing AI contracts
    • Licence keyValidated offline. Never contacts the hub.For restricted networks
    • On-premiseThe hub inside your perimeter.For air-gapped sites

    The paperwork

    Security questionnaire
    Completed on request, including sub-processors and data flows.
    DPA
    Available, and signed before any pilot touches production data.
    MSA and vulnerability SLA
    Standard terms, negotiable for company-wide rollouts.
    Audit trail
    SSO and SAML with an audit trail on the Enterprise plan.
    Compliance
    ISO 27001 certified. GDPR compliant. Made in Germany, Karlsruhe.
    Training
    Customer data and screenshots never train models, ours or anyone else's.
    Data flow
    Only the screenshot, the step text and the current phase go to the model. Reports, screenshots and secrets stay local.
    Models
    AskUI Hub, Anthropic, Azure AI Foundry, Anthropic-compatible gateways including vLLM, or self-hosted. OpenAI-compatible.
    Secrets
    OS-encrypted, write-only once saved, per-secret agent visibility, redacted from transcripts.
    Deployment
    Hub: AskUI-routed inference, online activation. Hub with your model: your provider, no inference fee. Licence key: offline activation, no hub contact. On-premise: your hosting, unmetered agents.
    Plans
    Scoped per organization and quoted in writing. How plans are scoped