Practice EU AI Act For law firms For private equity Work About Get in touch
Built for law firms

AI that gives your fee-earners
their hours back.

AI lands in a firm when three things fit together: the right platform, the workflows rebuilt around it, and the people who reach for it every day. I work across all three, independent and platform-agnostic, so the licences you already pay for turn into hours your fee-earners get back.

The state of play

The firms pulling ahead treat AI as an operating change, not a purchase.

The conversation in legal has moved on from whether to adopt. The live question is how to get from a few enthusiastic pilots to capability the whole firm runs on. That rarely stalls on the tool, whether you land on Harvey, Legora, or a build around your own process. It stalls when the platform, the redesigned workflow, and the people change one at a time instead of together, and the licences end up gathering dust.

Inside any firm, some people already pull far more from these tools than others. Closing that gap is what turns a handful of early adopters into firm-wide capacity. And capacity is the right frame. The reflex worry is that efficiency eats the billable hour. My read is the opposite: the same headcount takes on work it used to turn away, margins improve, and a smaller firm starts to compete with a larger one.

Use cases

AI use cases in legal work.

Matter intake and triage

New matters classified, conflicts surfaced, and routing decided before a fee-earner spends an hour on them.

First-pass document review

The first read across a large document set, so lawyers spend their hours on the judgment calls instead of the first pass.

Contract review and drafting

Review, drafting, and a firm-specific clause and precedent library, built around the way your teams actually work.

Litigation and eDiscovery

Search, classification and chronologies across large evidence sets, where the hours pile up and the deadlines are fixed.

IP and trademark

Watching, classifying and drafting around registers and filings, the rules-heavy work a process-specific build handles well.

Knowledge management

Your own precedents and know-how made findable, so the firm’s collective experience is one question away.

Confidentiality

Privilege is the line that does not move.

Use the strongest models on your most sensitive work, with the data posture decided matter by matter and the data-flow and contracts documented to back it up. Sanitise, process, and reinject, with governed agent access to the systems you already run.

Routine work

The platform’s standard zero-retention enterprise contract. Documented, and enough for the bulk of day-to-day matters.

Client or regulation-sensitive

EU data residency on top, with the data-flow and contractual trail a client or a regulator can inspect.

Crown-jewel matters

Run where nothing identifiable leaves the firm. On-premise, or in-firm de-identification.

How I work

Three ways in.

Start small and bounded, or hand over the whole programme. Either way, the firm keeps the controls.

01 Discovery Sprint

Two to three weeks, a fixed fee, to pressure-test the plan: whether the platform you are leaning toward is the right call, where the privilege posture needs work, and which workflow pays back first. You keep the output whether or not we go further.

You get A clear-eyed read and the order to build in.
02 Embedded delivery

I work alongside your team and take one use case the whole way. Redesign the workflow from zero, put the machine on the first pass and your lawyers on the judgment calls, build it, ship it, and stay until your people use it day to day.

You get A workflow live and adopted, not a prototype that stalls.
03 Interim or Fractional CAIO

I hold the AI programme with you. Strategy, the AI roadmap, build-versus-buy, governance, and standing up the capability inside the firm.

You get A function that runs without me when the engagement ends.
Who you would be working with

I rebuilt a PE-backed software company’s operations around AI in production, as Director of AI Transformation. I argued the case on AI regulation at the House of Lords and organised a debate on AI consciousness at Middle Temple Hall, in one of the Inns of Court, with Gary Marcus, Anil Seth and Marcus Hutter. For a firm weighing privilege, adoption and the EU AI Act in one decision, that is the seat I have already sat in.

The EU AI Act, for firms

Governed ahead of the deadline.

The Act’s high-risk obligations, the ones that reach recruitment and case-handling AI, moved to 2 December 2027 under the adopted Digital Omnibus, so a firm has real runway. Article 50 transparency duties still start on 2 August 2026, while the fines regime has applied since 2 August 2025. And the duty that already bites is Article 4: since February 2025 your fee-earners have to be AI-literate, which is the same enablement work that makes adoption stick. I classify your AI use, set the governance, and time the documentation to the rules, so the compliance comes out of the rollout instead of a separate scramble.

See the EU AI Act assessment →

Now let’s see
if we’re a fit.

A 30-minute working call. I talk about where your firm is with AI and where the value would land first. I work alongside your counsel; the legal calls stay with them.

Or grab the EU AI Act worksheet →

Built and run by a practitioner who shipped an enterprise AI transformation end to end and argued the case on AI regulation at the House of Lords.

More on who delivers it →