How pricing works

Flat monthly. No per-token surprises.

One predictable fee covers the private AI, the hosting under it, and the people operating it — on dedicated European hardware, not a shared public cloud. Billing is in EUR, net of VAT. You always get a written quote before you commit — and you can always leave.

Predictable, not per-token

A fixed monthly fee with a generous included quota — not a bill that moves with every prompt. We do measure usage, and if you consistently run past your tier we talk first.

A number in writing

Managed production runs under a written SLA — 99.9% availability, RTO/RPO defined per plan. In the contract, not the footer.

No lock-in

Open components, your code and data, documented export, defined exit. How ownership works →

The platform

Private AI & hosting, operated

from €500 / month

Your own private AI plus the full managed stack under it — hosted, monitored and operated for you on dedicated EU hardware.

  • Open-weight models served privately — your prompts never train anyone else’s AI
  • A private knowledge base (RAG) on your own documents
  • Managed hosting bundle: web, PostgreSQL, S3-compatible storage, email, DNS
  • Monitoring, backups, security updates — we operate it
  • Written SLA on production — availability, RTO and RPO in the contract
Talk to us →

AI-assisted delivery

Build & run

from €1,500 / month · 6-month engagement

We build the app for your one confidential workflow — then ship it through governed releases and run it.

  • Everything in the platform tier, for the app we build you
  • ATAILA Coder builds it with you — senior architect reviews every step
  • Human-gated, audited DEV → UAT → PROD releases
  • Security review before go-live; ongoing right-sizing after
  • One-time onboarding priced with the quote
Bring us a workflow →

Licensed & self-hosted

Enterprise — your hardware

On request priced per engagement

License the platform and run a single-tenant instance in your own datacentre — operated by us, or by your team.

  • Single-tenant: your jurisdiction, your servers, your capacity
  • No shared-hardware limits — the ceiling is the hardware you own
  • Operated by us, or your team trained until they run it confidently
  • Updates, security fixes and health monitoring either way
Compare editions →

Prices are net of VAT and are tier floors ("from") — the final monthly fee depends on your configuration and is confirmed in a written quote.

What the flat fee replaces

Private AI is usually sold three other ways. Each one leaves you holding something.

Buying a box and running it yourself

A capable AI workstation is €5–8k up front — then patching, backups, GPU failures and depreciation are yours, and nobody answers at 3 a.m.

Ask: who operates it in year two?

Having AI installed on your own server

A setup fee plus a monthly fee to remote-manage software on hardware you still have to buy, host and power — no redundancy, no hosting behind it.

Ask: what infrastructure stands behind the SLA?

Paying per token in the public cloud

Public AI APIs run €300–1,000+ a month at company scale, the bill moves with every prompt — and your data leaves your control.

Ask: where do the prompts, logs and embeddings live?

The ATAILA answer: one flat fee, the AI and the hosting on redundant EU infrastructure, operated by a team that answers — with the SLA in writing.

Straight answers

Why "from" prices?
Because configurations genuinely differ — models, storage, environments, integrations. The anchors above are real floors, not bait: you get a written, itemised quote before you commit, and the monthly fee is flat once agreed.
What does the SLA actually say?
Managed production carries a written availability commitment of 99.9%, with recovery time (RTO) and recovery point (RPO) defined per plan — in the contract, not in a footer. The stack behind it is redundant and multi-site, monitored around the clock.
Is there a usage limit? What happens if we exceed it?
Yes — and we’d rather be straight about it than surprise you. Each tier includes a generous quota sized to normal business use, and we measure consumption. If you consistently run past it we contact you and move you to the right tier (or a dedicated node) — we don’t silently meter you, and we don’t let one tenant’s heavy usage degrade everyone else sharing the hardware. Sustained heavy inference is a real cost on real GPUs; the fair-use limit is what keeps the flat fee honest for everyone.
Is per-token ever the better deal?
Honestly: yes — if your team touches AI a few times a week, a public subscription is cheaper and we’ll tell you so. A private platform pays off when AI is part of the daily workflow, or when the data must not leave.
Can we leave?
Any time. Everything is built on open components; your code, data and files are yours, with a documented export on demand and a defined exit.
Are these prices final?
They are net of VAT and they are anchors — the floor of each tier. Every engagement is confirmed with a written quote before any commitment.

Want a number for your case?

Tell us the workflow and the constraints — you'll get a written, itemised quote, not a sales runaround.