A4 by Cognethics pricing

The same platform at every tier. What changes is how many Studio seats and compute units you start with, how your people’s agents are isolated, and what we sign up to.

In your Studio, separate from the application itself — walled off from our source code, from the platform database and from other customers. Your organization’s Studio runs on its own virtual server, separated from every other customer by hardware virtualization; how your own people are separated from each other is the isolation option you choose.

Each compute unit lets your agents use up to 2 vCPU and 4 GiB of memory at once. Each seat includes one; add more in whole units for $1,200 a year. On Collaborative and Shared, your seats’ units are pooled.

How much your own people’s agents can see of each other. Collaborative and Shared workspaces work side by side, sharing compute. Zero-Trust gives every named person a dedicated virtual server — separate processes, files, memory and kernel — which is what an auditor expects when different staff hold different lawful access to the same records. It is an add-on on Studio and Team; on Enterprise it is scoped to the seats that need it. Because each Zero-Trust seat is its own virtual server — in effect an extra Studio — an extra Zero-Trust seat is $3,500 a year, against $1,500 for Collaborative and $1,900 for Shared, which share one Studio’s resources.

Your Studio agents use your own model account and you pay your provider directly, at your rate. The AI built into the platform itself is included in your plan. We never charge for inference, add no markup and take no cut.

No. Every limit is a count you can see — Studio seats, compute units, storage blocks — and none of it is billed by consumption. If you need more, you add a seat, a compute unit or a block. Nothing changes mid-term.

Neither. Users and Studio seats are different things. A user is anyone who signs in to the platform — Studio includes up to 100, Team up to 1,000 and Enterprise is unlimited. A Studio seat is a workspace for someone who builds and runs agents; Studio includes one, Team four, and you can add more from $1,500 a year. Agents are never a line on the bill, on any tier.

Studio starts with a one-hour kickoff at no charge, and you can add hands-on help in prepaid blocks of hours. Team and Enterprise onboarding is real, scoped work, quoted with your subscription so you know the whole first-year number before you sign.

Yes, at any point, prorated to the remainder of the term. Nothing you have already paid for gets switched off.

Talk to us. Enterprise is sized to the Studio seats, compute, storage and compliance your work needs, and it comes back as one annual number before you sign.

All plans are an annual fee, invoiced in advance, with quarterly invoicing available on Enterprise. Onboarding is billed half on signature and half at go-live. Terms are 12 months, renewing automatically, with 60 days’ notice either way and 90 days’ notice before any change to the fee.

In AWS in the United States. Which parts of the platform run on our own infrastructure and which call an external model provider is set out on our trust page. Read the trust page →