Skip to content
Book a proof run

Live demos

Umpire · Verity

See it work before you buy it.

Proof before purchase starts here. Each demo runs in your browser and contacts no external system. Open one full screen, or see it next to the product it belongs to.

  • Umpire

    Guardian: every action checked

    Run five agent actions through the prepare → mandate → token → submit protocol. Watch one go to a human for approval, one get refused for its content, and a payment get blocked because its payload changed after approval. Freeze the agent, then tamper with a receipt and see the chain catch it.

    Design simulation. Synthetic actions; hashes are computed live in the page and no external system is contacted.

  • Umpire

    Agent Doctor: the watch

    Six agents measured against their own seven-day baselines. Pick one to see its vitals, a diagnosis and a prescription — and notice that every fix is a proposal to a person, never an action on its own.

    Design simulation. All agents, numbers and baselines are synthetic.

  • Umpire

    AgentCert: the test

    What an AgentCert attestation contains, how anyone can verify it without trusting the issuer, and two ways an insurer uses the grade — pricing AI-agent liability cover and routing claims filed by agents.

    Design simulation. Grades, providers, premiums and lanes are illustrative.

  • Verity

    Verity: one renewal, start to finish

    An account manager connects five apps and works one customer renewal from Monday morning to the moment something goes wrong — and gets fixed. Use the arrow keys or the buttons to move through the scenes.

    Fictional companies and people.

After the demo

Proof before purchase

Four steps from a first look to production.

  1. Watch it workThese demos, or a 30-minute live session on a domain that isn’t yours.
  2. Your questionsBring real questions, the actions you worry about an agent taking, or the list of apps you run. We work through them live, under NDA.
  3. A sandbox on your materialOne narrow use case on a masked slice of your own material, in an isolated environment, with success criteria agreed in writing first. Scoped and paid, typically two to four weeks.
  4. Production pilotWired into your real systems for a defined group of users, and measured against the sandbox criteria under live volume.
Where we draw the line

At every step you see what each capability does and watch it happen. Model weights, rulebook logic, knowledge-graph schema and connector architecture stay closed; what you can license, and on what terms, is a contract conversation.