One private AI chat for your whole team
The BrainOutput platform is the core of everything we deliver: a chat interface, memory per user, model routing by confidentiality, agents for long-running work and administration — deployed where your data must live.
What the platform includes
Everything your team needs to work with AI, in one place:
- One chat interface for every user, with conversation history and shared spaces.
- Memory per user, so the assistant keeps context across conversations.
- Model routing by confidentiality, applied to every request.
- Agents that carry out long-running work under human review.
- Administration and policy controls for your whole organisation.
On your own hardware
We supply and provision machines sized to the models you need — desktop AI units, RTX workstations and large-memory NVIDIA systems in the ~750 GB memory class. Exact models and specifications are confirmed per engagement.
On your hardware, prompts and documents never leave your premises: inference is local, and the only outbound traffic is what your policy allows.
Or in a private cloud
The same platform, deployed on dedicated private-cloud infrastructure that we operate for you — no shared public tenancy, and the same routing policy as on-premises.
You can start in a private cloud and move to your own hardware later; the platform and your policy move with you.
Model routing by confidentiality
Every request is classed as sensitive or not sensitive. Sensitive requests — client files, contracts, accounts, health or guest data — are answered by a private model on your infrastructure, full stop.
Non-sensitive requests may be routed to a cheaper or stronger cloud model, but only when your policy allows it. You set the policy; the platform enforces it.
Your sensitive work never leaves; the rest goes wherever is cheapest or best — only when you say so. On a new installation everything stays private; a cloud model is used only if you enable it.
LLM-agnostic by design
BrainOutput works with models from any provider — open-weight or commercial — and routes each task to the model that fits its confidentiality, cost and difficulty.
Bring your own API keys, run local models, or mix both. When a better model appears, you switch the routing, not the way your team works.
Add-ons run from the same chat
Every add-on is used from the platform's chat — one interface, one policy, one audit trail.
DocMatch
Document analysis. Designed for law firms (cases and disclosure) and accountants (matching receipts, invoices and bank statements).
Receptionist
An AI phone receptionist for any business — every call answered in your name.
Companion
A voice and screen assistant across PC and phone.
Presence
Presence and wellbeing through the camera. Frames never leave your site; presence only, never identity — nudges, not health advice.
Email & message management
Triage, labels, reply drafts and auto-replies across email and messaging.
Frequently asked questions
What is private AI?
Private AI is AI that runs on infrastructure you control — your own hardware or a dedicated private cloud — instead of a provider's shared public service. BrainOutput is a private AI platform: one chat interface where your team asks for work and gets it done, with your confidentiality policy applied to every request.
Does our data leave our premises?
Sensitive work never does: it is answered by a private model on your hardware or in your private cloud. Non-sensitive work may use a cloud model only when your policy explicitly allows it. On a new installation everything stays private; a cloud model is used only if you enable it.
What exactly happens to a sensitive request?
It is classed as sensitive, routed to a private model running on your hardware or in your private cloud, and never sent to an external provider. The classification and routing rules are yours to set.
Which AI models does BrainOutput work with?
BrainOutput is LLM-agnostic: it works with open-weight models running locally and with commercial cloud models, and routes each task according to its confidentiality, cost and difficulty. When a better model appears, you change the routing, not the way your team works.
Can we keep our existing AI provider?
Yes. Bring your own API keys and BrainOutput routes to your provider's models alongside private ones, under the same policy.
Which languages does BrainOutput support?
We work with organisations across the European Union, and this website is published in the 24 official EU languages. Your team uses the chat in its own language; how well each language is handled depends on the model, and we check it for your languages during the AI diagnostic.
What hardware do we need?
That depends on the models you want to run. We size and supply the hardware — from desktop AI units to large-memory NVIDIA systems — and provision it with a working model stack. You can also skip hardware entirely and run in a private cloud we operate for you.
Can we start in a private cloud and move on-premises later?
Yes. It is the same platform in both deployments, so moving is a migration of infrastructure, not of tools or habits.
How is this different from a public AI chat subscription?
A public subscription sends your prompts to a provider's shared service. BrainOutput runs where your data must live, works with any model, applies your confidentiality policy to every request, and adds capabilities — document analysis, phone reception, email triage — that act inside your organisation.
What add-ons are available?
Five add-ons run from the same chat: DocMatch (document analysis), Receptionist (AI phone receptionist), Companion (voice and screen assistant across PC and phone), Presence (camera-based presence and wellbeing, never identity) and Email & message management. They share one interface, one policy and one audit trail.
How do we get started?
Book a demo to see the platform on demo data, or start with an AI diagnostic: a structured assessment of your processes that ends with a prioritised plan, yours to keep. We do not publish prices; every engagement is scoped and quoted individually.
See the platform on your use case
Book a demo: we will show you the chat, the routing and the add-ons on demo data, and discuss your constraints.