No. Most firms start with AI inside their own Microsoft environment, which is already private to you. A private, in-house model is there for the most sensitive work, and we only suggest it where the data or your regulator calls for it.
AI that works for your firm, and keeps your data where it belongs.
In law, finance and healthcare, artificial intelligence can take real work off your team: drafting, review, research, the routine that fills the day. The catch is the data you hold, which is exactly the data that must never end up in the wrong tool. We help you get the benefit and keep control, from AI inside your Microsoft environment to a private model that never leaves your four walls, all governed to ISO 42001 so it stands up to your regulator and your clients.
Real work, taken off your team's plate.
AI for law firms
Draft first-cut contracts and letters, summarise long bundles, run first-pass document and disclosure review, and speed up research, without a matter leaving your control.
AI for accountancy firms
Draft client reports and commentary, prepare and sense-check working papers, summarise regulatory updates and handle routine month-end prep, with client data kept inside your controls.
AI for healthcare
Draft letters and clinical admin, summarise notes and triage correspondence, and free your people for care, without patient data ever being handled unsafely.
We already look after legal and accountancy firms day to day (see our IT support for legal and accountancy firms), and our practical guide to Copilot and AI for fee earners is a good next read.
Not all AI is equal. The more sensitive your data, the more control you need.
"AI" runs from free tools anyone can open in a browser to a model that lives entirely inside your own walls. Think of it as a traffic light: the further your work sits towards sensitive and regulated, the further along the spectrum your AI needs to sit. Part of what we do is place each task in the right spot, and give you the option at every point.
Free public AI.
Consumer tools like the free version of ChatGPT. Fine for general, public-facing tasks. The risk is the moment someone pastes a client name, a matter or a patient detail, because it can leave your control for good.
Approved tools, governed.
Tools like Claude under a data agreement and clear rules on what may go in. Safe for internal, non-sensitive work, with a record of how it is used.
AI inside your Microsoft environment.
Microsoft Copilot works on your data inside your own Microsoft environment, so it never leaves the space you already control. The right home for most sensitive day-to-day work.
Your own private AI.
A model that runs entirely inside your environment. Nothing is sent anywhere. The strongest answer there is to a regulator or a client who asks where your data goes.
For the work that can never leave your four walls: private, in-house AI.
Where the data is too sensitive to send anywhere, or a regulator or a client demands it, we can set up AI that runs entirely inside your environment. Your data stays put, nothing goes to a third party, and your team still gets the benefit. It is the clearest answer you can give to "where does our data go": nowhere. We scope it to what your firm actually needs and build it with you, so it becomes a capability your firm owns. The infrastructure it runs on is home ground too: designing and managing it is our IT infrastructure and network management service.
Governed to ISO 42001, built on twenty years of guiding firms through change.
Wherever you sit on the spectrum, the governance is the same: clear rules, corrected permissions, and the evidence to prove it. Because we work to ISO 42001 and run the same system on ourselves first, your AI use is governed and defensible, not improvised.
- Clear rules for what may go into which tool, written for your people, not technologists.
- Permissions corrected so AI can only reach what each person should reach.
- A policy, a tool register and an incident route you can show in a client, audit or regulatory review.
Common questions
Do we have to host AI ourselves?
Can our people use AI on sensitive client or patient data at all?
How do we answer our regulator or an auditor on AI?
The first step is a conversation, not a contract.
Book a free AI Discovery Call. We will show you, on your firm's own world, where AI would help, where on the spectrum each task belongs, and how to adopt it without putting confidential, client or patient data at risk.
Book a meeting with our AI leadDave Coleman, EPX IT's AI leadContact us
Keep exploring the AI hub.
The AI hub home
Putting AI to work in your business, safely
Where do you stand?
Take the free two-minute AI readiness quiz
AI Foundations
Your AI policy, rules and guard rails
Microsoft Copilot
Set up properly and actually used
Prompts and use cases
Ready-to-use Copilot prompts for real jobs
AI Assured
Monthly monitoring for shadow AI and oversharing
How we use AI ourselves
Nearly 90 hours a month back, measured on our own business
