Technology & AI Advisory
for Professional Services Firms
Accountancy practices, law firms, consultancies and agencies. Businesses where the product is your people's judgement, the constraint is confidentiality, and the pricing model quietly punishes you for getting faster.
Why This Sector Gets Different Advice
Generic AI advice assumes a business that makes or sells something. A professional services firm sells hours of expert attention, holds client material under obligations that predate the internet, and makes decisions through partners rather than an executive. Each of those changes the answer.
The technology questions are not especially exotic. The commercial and professional constraints around them are, and that is where most firms get stuck.
The Billable Hour Problem
Worth confronting first, because it decides whether any of this is worth doing.
If you bill by time and a tool makes a task three times faster, you have just reduced your own revenue. Every firm we speak to arrives at this and most stop there, which is a shame, because there are only three sensible responses and all of them are workable.
Point it at non-billable work first
Proposals, internal research, know-how, onboarding, the administrative tail that surrounds fee-earning work. The saving is pure and nobody has to renegotiate anything. This is where we start with most firms and it is usually enough to justify the exercise on its own.
Take the capacity rather than the saving
Same team, more matters, without hiring. Requires that you have demand you are currently turning away or slow-walking, which is worth establishing honestly before building a plan on it.
Change what you charge for
Fixed fees on the work that has become predictable. The strategic answer, the slowest, and the one that needs partner agreement rather than a software decision. Firms that get there first tend to take the commoditising work off everyone else.
Any adviser who skips this conversation and goes straight to tooling is selling you software, not advice.
Where AI Genuinely Helps a Firm
- ✔ First-pass document review. Reading long documents and telling you where to look. The saving is in the reading, not the judging.
- ✔ Drafting from your own precedents. Assembling a first version from material your firm has already produced and approved, rather than from the open internet.
- ✔ Making your own know-how findable. Most firms have twenty years of excellent work nobody can locate. Answering internal questions from your own document store is among the highest-value uses available and the least discussed.
- ✔ Proposals and tenders. First drafts from previous submissions and case studies. Often the fastest measurable win in the firm.
- ✔ Client correspondence and file notes. Drafts for a fee earner to approve, and turning meeting recordings into notes and actions.
- ✔ The administrative tail. Time narratives, engagement letters, conflict checks paperwork, the small tasks that consume senior people in aggregate.
And where it does not: professional judgement, advice that carries your firm's liability, anything a client would rely on without a person having read it, and any calculation going to a regulator or a court. The pattern is the same as everywhere else - AI drafts, a qualified human decides - but the consequences of getting it wrong are steeper here. Our general breakdown by function is on what AI can actually do for your business.
Confidentiality Is the Real Constraint
Client material under privilege, commercially sensitive files, personal data held under a professional duty, engagement terms that specify where information may be processed. For a lot of firms this is the reason nothing has happened yet, and it is a legitimate reason rather than an excuse.
It is also solvable. The route is to establish what each tool does with your content, sanction one that meets the standard, write the rule down, and keep the genuinely privileged material out of anything you do not control. Is it safe to put company data into AI sets out the questions to put to vendors, and our governance policy guide has the written template.
Where the material must not leave your network at all, models that run on your own hardware are a genuine option now. The full comparison, including what you give up, is in local AI versus cloud AI. Firms with a small quantity of genuinely privileged work and a large quantity of ordinary work usually end up running both.
Expect your own clients to start asking. Corporate clients are adding AI questions to supplier questionnaires, and a firm that can produce a dated policy answers in a minute what others answer in a fortnight.
The Systems Underneath
A recurring finding: the firm's ambition is blocked not by AI but by the practice management system installed in 2011 that will not let data out, or by twenty years of documents in a shared drive with no structure. AI applied on top of that produces confident nonsense, because the material it is reading is a mess.
When that is the situation we say so, and the work becomes modernisation rather than AI. When a codebase becomes a business risk covers how to judge the point at which an old system stops being merely annoying. If the question is whether to replace a system with something bought or something built, the build vs buy framework applies, and for most firms in this sector the answer is buy.
How We Work With Firms
Opportunity review
Two to three weeks. Where the non-billable hours go, which tasks are worth testing, what each is worth, and the honest list of what to leave alone. Fixed scope.
Confidentiality and policy
A written rule your partners can approve, a sanctioned toolset that meets your obligations, and a briefing for the whole firm.
Partner and board briefings
A session for people who need to make a decision rather than run a project. No demos, no vendor in the room.
Ongoing advisory
A fractional Chief AI Officer or fractional CTO arrangement of a day or two a week while the change beds in.
If your firm advises on transactions, we also sit on the other side of that work: independent technical due diligence for corporate finance teams and their clients, including how a target's own AI usage and data handling stand up to scrutiny.
Local to the Thames Valley
We are based in Berkshire and work in person across the M4 corridor, which suits partner meetings that are considerably more productive in a room. See Thames Valley, Reading and Berkshire, or Hampshire. Firms further afield we work with remotely, which most now prefer for anything short of a partnership decision.
Other sectors we publish on: financial services and fintech, where the regulatory perimeter rather than professional privilege sets the constraints.
Run a Firm and Tired of Being Sold To?
A free half-hour with someone who will tell you which two or three things are worth doing, and which are not worth your partners' time.