Simple seat pricing. You choose where the AI runs.

You pay per seat per month, billed monthly or once a year through secure checkout. Self-serve plans start at 3 seats for every setup and go up to 100 seats.

Yes, any time from your dashboard. Adding or removing seats is prorated on your next invoice.

Annual billing gives you 2 months free compared with paying monthly, billed once a year.

Prices exclude VAT. VAT is added at checkout where it applies.

Yes, cancel any time from your dashboard. Your seats keep working until the end of the period you have paid for.

No. With your current AI you keep your existing ChatGPT, Claude or Codex plan. With AI hosted in your country, you pay the provider directly on your own key. On your own servers, the models run on your hardware. The configurator shows an estimate for each, so there are no surprises.

Yes. Every plan starts with a 7-day free trial. You add a card at checkout, nothing is charged today, and the first charge is on day 8. Cancel before then and you pay nothing.

No. Recent Apple Silicon Macs (M1 or later) can run small open models directly on each person's machine, because the Mac's unified memory doubles as graphics memory. At 4-bit, a 7 to 8B model such as Qwen 3 8B, Llama 3.1 8B or Mistral 7B is about a 5 GB download and runs well with 8 to 16 GB of memory (16 GB is comfortable). A 14B model needs around 16 to 24 GB, and a 32B model needs 32 GB or more. Prompts never leave the laptop. A shared GPU server only pays off when many people need a larger model at the same time. The local model guide shows the setup.

Yes. Individual plans for one person live on the bots page.

Email us and we set it up with you. We reply within one working day.