How it works
A capable AI system, running inside your office
No accounts with overseas providers. No per-message charges. No transmission. Your team gets the tool they were reaching for anyway — it just happens to be sitting in the comms cupboard.
In one paragraph
A dedicated machine lives on your network and runs open AI models locally. Staff open a page in their browser and use it exactly as they would use any AI chat tool. Because the model is running on that machine, the document never travels anywhere — there is no request to an external service, because there is no external service. Unplug the internet and it keeps working.
What it does well, and what it doesn’t
Any vendor who gives you only the first list is selling you a disappointment. Both lists matter.
Genuinely good at
- Summarise long documents, transcripts and bundles
- Draft correspondence, notes and reports from your own material
- Answer questions across your internal files and precedents
- Extract structured data from unstructured documents
- Rewrite and reformat text for different audiences
- Assist with code, spreadsheets and calculation logic
- Work entirely offline, including at remote sites
Will not do
- Match the very largest commercial models on novel, frontier-level reasoning
- Run the absolute largest open models at full speed — some run slowly, some not at all
- Browse the live internet (by design — that is the point)
- Replace professional judgement or remove your duty to check its output
- Guarantee it will never be wrong — every AI system gets things wrong
- Make you compliant on its own; it removes one specific risk, not all of them
From first call to running system
- 01
Assessment
We work out what your team is actually doing with AI now, what it would need to do to be worth replacing, and what you are contractually or professionally obliged to protect. This is also where we tell you if you would be better off with a policy and a subscription. That happens, and we say so.
- 02
Scoping
Number of concurrent users, the kind of documents involved, whether it needs to reach your existing systems, and where it will physically live. This determines the specification. We size it to the workload rather than selling the biggest thing available.
- 03
Build and configuration
The system is assembled, the models selected and installed, and the interface configured before it arrives. We test it against your actual use cases — your document types, your questions — rather than a generic benchmark.
- 04
Installation
We install it on your network, alongside your existing IT provider if you have one. Your team reaches it through a browser. There is nothing to install on individual machines and no accounts with any external service to create.
- 05
Handover and training
A short session per team on what it does well, what it does badly, and how to check its output. The failure mode we most want to avoid is people trusting it more than they should.
- 06
Ongoing support
Hardware support, model updates as better open models are released, and someone to call. The models improve over time and the system improves with them — without the specification changing.
On cost
Why there is no price on this page
Because a number without a scope is meaningless, and we would rather not pretend otherwise. A system for four people doing document summarisation and a system for thirty across multiple offices are not the same purchase, and quoting the smaller one to make a page look attractive would waste your time and ours.
What we will tell you plainly: it is a capital purchase plus a support arrangement, not a per-seat subscription. It does not increase when you hire, and it does not increase when your team uses it more — which matters, because a tool that costs money per message quietly teaches people to use the free one instead.
You will get a written scope and a fixed figure before you commit to anything. The assessment conversation is free and carries no obligation.
Practical questions
- What does the team actually see day to day?
- A chat page in their browser, on your network. Type a question, paste or upload a document, get an answer. If they have used any AI chat tool, there is essentially nothing new to learn.
- Where does it physically go?
- A server cupboard, a comms rack, or a shelf in a locked room. It is a compact unit, roughly the size of a small desktop, and it runs quietly on a normal power point. It needs a network connection to your office — not to the internet.
- What happens when better models are released?
- The open model ecosystem moves quickly, and new releases can be installed on the same hardware. That is one of the genuine advantages of this approach: capability improves over time without a new purchase, within the limits of what the hardware can hold.
- Can it read our existing files?
- It can be configured to search and answer questions across a document library you point it at. How cleanly that works depends on where your documents live and what that system exposes. We establish that during scoping rather than promising it up front and discovering the integration does not exist.
- What if the hardware fails?
- Support and warranty terms are part of what we scope — including how quickly a failed unit gets attended to, which matters more for some firms than others. We will be specific about this in writing before you commit, because a vague answer here is worth nothing.
- Do you lock us into anything proprietary?
- No. The models are open and the software is standard. If you ever want to take it in-house, move it, or have another provider support it, you can. We would rather earn the renewal than trap you into it.
No obligation
Find out whether this fits your firm
Tell us how your team is using AI today and what you cannot afford to have leave the building. We will tell you honestly whether an on-premise system is the right answer — including when it is not.