- Home
- AI consultancy
- Private AI
Private AI for companies: a ChatGPT-style assistant that keeps your data in-house
Your team gets the AI tools they want, and your client files, contracts and designs stay under your control. We set up private AI on your own hardware, in your own cloud or on infrastructure we run for you, for businesses anywhere in the UK.
Why companies want private AI
Most staff already use AI. The worry is what they paste into it: client details, contracts, pricing, designs. With free tools, that information sits on someone else’s servers under someone else’s terms.
Private AI gives your team the same kind of assistant without that risk. You decide where it runs, who can use it and what it can see, and nothing is used to train anyone else’s model.
If you’re weighing up ChatGPT itself, our guide Is ChatGPT GDPR compliant? explains the differences between its plans.
Where it runs
We pick the setup that fits your data, your team and your budget.
In your office
A dedicated AI server on your premises, such as a Mac Studio or a GPU server. We can supply the hardware, set it up and support it.
In your cloud
Deployed into your own Microsoft Azure or AWS account, in a UK region if you need your data kept in the UK.
On infrastructure we run
Hosted for you on AWS, Azure or a private server, run and secured by us, in a UK region where needed.
Hybrid
Sensitive work stays private, and routine work goes to a business cloud account whose terms keep your data out of model training.
What you get
Built with proven open-source tools, such as Ollama and Open WebUI, and open-weight models chosen for your work and checked for licence terms.
- A ChatGPT-style assistantYour staff sign in with their work Microsoft or Google accounts and chat as they would with ChatGPT.
- Chat with your documentsAnswers from your own shared drives and files, with the source shown and existing permissions respected.
- Private transcription and notesMeetings and calls summarised without recordings going to an outside service.
- Drafting from your templatesLetters, reports and replies written in your format and tone.
- Private agentsThe same AI agents we build for routine work, running on your private setup.
Local models vs ChatGPT: the honest version
Open models running on your own hardware are good at the everyday work most businesses need: summarising documents, answering questions from your files, drafting emails and reports, and pulling details out of forms.
They aren’t yet as strong as the best cloud models on the hardest jobs, such as long, complex reasoning or very polished writing. We’ll be straight with you about where that line sits for your work.
That’s why a hybrid setup often works best. Client and confidential work stays on the private system, and general tasks go to a business cloud account with the right terms in place. You get privacy where it matters and the strongest models where it doesn’t.
Who it’s for
Any business that holds information it can’t afford to share. These are the firms we hear from most.
- Solicitors and conveyancersClient confidentiality and privilege.
- Accountants and bookkeepersClient finances and tax records.
- Financial advisersClient circumstances and suitability reports.
- Dental and healthcare practicesPatient records and clinical notes.
- HR and recruitmentCVs and personal details.
- Manufacturers and engineersDrawings, designs and customer IP.
What’s included
Everything needed to run it properly, not just to switch it on.
- SizingHardware or cloud sized from how many people will use it, and for what.
- Setup and sign-inInstalled, connected to your documents and set up with your work accounts.
- Access controlTeams only see what they should, and admins control which models people can use.
- Logging and backupsA record of use for your peace of mind, and backups you can restore from.
- Updates and new modelsSecurity updates applied and new models tested before they go live, on an AI care plan.
Built to be trusted
We’ve delivered Kubernetes, Chef, Linux infrastructure and automation work for UK government agencies. We bring the same habits to AI: automate it, test it, document it and secure it.
We use AI agents in our own business, with a person approving every step, and we host and secure infrastructure ourselves. See our managed hosting.
We’re independent. We aren’t a partner or reseller of OpenAI, Anthropic or Microsoft, so we recommend what fits your business.
- A person approves what mattersAnything that goes to a customer, spends money or changes your records waits for someone to say yes.
- Every action is loggedYou can see what the agent did, when it did it and why.
- Only the access it needsAgents get the narrowest permissions that let them do the job.
- Off at any timeYou can switch any agent off straight away, and your team carries on as before.
- Your data isn’t used for trainingWe set up business accounts or private models so your data isn’t used to train AI models, and we show you the terms.
- You own itThe prompts, workflows and code are yours, documented and handed over.
What happens when you get in touch
No sales script and no obligation. We’ll tell you honestly whether AI is the right answer for the job.
- We reply within one working daySend a few lines about your business and what you’d like to change.
- A free 30-minute callWe talk through the work, what AI could do and what it shouldn’t.
- A written scope and a fixed quoteBefore any work starts, so you know exactly what you’re getting.
Private AI questions
Private AI or public AI: what’s the difference?
Public AI tools run on the provider’s servers under the provider’s terms, shared with millions of other users. Private AI runs on hardware or cloud accounts that you control, so you decide where the data goes, who can see it and how long it’s kept.
Should we run AI on-premise or in the cloud?
On-premise suits firms that want data to stay in the building and have a steady number of users. A private cloud suits teams spread across sites, or work that needs more computing power some days than others. Many businesses end up with a mix, and we help you choose.
How good are local models compared with ChatGPT?
For everyday work such as summarising, drafting, searching documents and answering questions from your own files, current open models do a good job. For the hardest reasoning and writing tasks, the best cloud models are still ahead. That’s why we often suggest a hybrid: private for sensitive work, a business cloud account for the rest.
What hardware do we need, and what does it cost to run?
It depends on how many people will use it and what for. A small team can often run well on a single Mac Studio or a server with one good graphics card, while larger teams need more. We size it from your real usage and give you a fixed quote for the hardware, setup and support, so the running costs are clear before you commit.
Can it search our SharePoint or shared drive?
Yes. We connect it to SharePoint, OneDrive, Google Drive or a file server, and it respects the permissions you already have, so people only get answers from documents they’re allowed to open.
Does any data leave the building?
Not with an on-premise setup, unless you choose a hybrid where some work goes to a cloud model. In that case we agree exactly which kinds of work can leave, and everything else stays on your hardware.
Can you set up a private AI note taker?
Yes. Meetings and calls can be transcribed and summarised on your own hardware or private cloud, so recordings never go to a third-party note-taking service.
Who maintains it?
We do, on an AI care plan: monitoring, security updates, backups and new models tested before they go live. If you’d rather run it yourselves, we hand over full documentation and train your team.
Talk to us about private AI
Tell us what kind of information your team works with and how many people would use it. We reply within one working day.