Fixed fee · Four days · A plan you own
A four-day, fixed-fee review for accounting, legal and financial advice practices. You leave with a ranked plan your business owns.
Licences, tools, add-ons, a pilot or two. The invoices arrive every month; the improvement doesn't. Nobody is sure whether to spend more, stop, or start again somewhere else.
That's the moment to call. Not because you need another tool, but because the question has changed from “which AI” to “where does this actually fit in a practice like ours, and what's stopping it.”
It isn't the technology.
Large organisation or ten-person practice, AI stalls for the same structural reasons. In a practice, a stall is the good outcome. The bad one is AI that half-works, with its mistakes going out under your name and your licence.
Last year the safe bet was a Copilot licence for everyone. This year half the team quietly uses ChatGPT or Claude, and which model leads changes every few months. A practice that built its workflow on one vendor's product is now stuck, or paying twice. The question was never which model is best this quarter. It's whether you can change your mind without losing your knowledge or your process.
The senior accountant knows how a trust distribution file is actually prepared. The senior associate knows which precedent to start from and why. The paraplanner knows how a complex rollover is really put together. The process doc says something else. An agent built on the doc produces confident, wrong output — and it looks fine until it isn't.
The agent goes live and nobody tells it when it's wrong. Output quality doesn't come from the model; it comes from people reviewing and correcting the work on a regular cadence, for months. Without that loop, the output stays mediocre and everyone decides AI doesn't work here.
The automation handles 90% of matters. The other 10% land nowhere, and the whole thing quietly gets abandoned.
A workflow gets built, then someone asks what the obligations are — the licensee's, the TPB's, the Law Society's — what has to go past a human, and what happens on an exception. Retrofitting that is more expensive than designing it in.
The AI project gets funded for tax, or for litigation, or for advice. The workflow it needs to improve runs across the whole practice — the practice management system, the document system, email, a shared drive, someone's head. The silo outside the budget is the roadblock, and the project is stillborn before anyone notices.
The review exists to find which of these is true for you, in which workflows, before any more money goes out.
Partners, principals and practice managers of accounting, legal and financial advice practices — typically 5 to 50 people — who already have AI in the business and want a defensible answer on what to do next.
Below five people there's rarely a cross-team problem to find. Above fifty you probably have an IT manager and a procurement process, and the conversation is different. In between is where a four-day review pays for itself.
It is not for practices that haven't started. If that's you, the first step is a conversation, not a review.
Firm size matters less than this. Not curious, not handing it to whoever likes tech: all in. In practice that means:
Everything below is yours. Take it to your own team, to us, or to another provider.
Readiness summary. The strengths, gaps and dependencies in the two or three workflows we reviewed — including the work that has to happen before automation is sensible.
Ranked opportunity list. Each candidate scored on business value, feasibility, data readiness, professional-obligation exposure and who would own it. Some will turn out to need a cleaner process or a better system connection, not AI. We'll say so.
Recommended first workflow with a plain explanation of the options and trade-offs, and what to fix first.
Indicative effort and cost for that first workflow: your team's time, software, specialist help, elapsed time — with the assumptions written down so you can challenge them.
90-day action plan. Named owners, preparation steps, decision points and the measures that would tell you it worked.
12-month direction. The sequence after the first workflow, so you can decide what to fund and what to defer.
Switching-cost view. For every recommendation: what you'd own outright (your knowledge, your process documentation, your data), what would depend on a particular vendor or model, and what it would cost to move. If you're already down a track that's hard to leave, we say what leaving would take — and sometimes the honest answer is stay.
Delivered as a written report and walked through with you live. We don't email it and hope.
Four working days of ours, spread over about three weeks of yours.
On site or remote. We sit with the people who do the work in the two or three workflows you've nominated. Systems, data, hand-offs, exceptions, what gets copied by hand.
From you
A sponsor, the process owners for half a day each, and agreement on what can be shown to us.
We map what we found against the failure patterns above, score the opportunities, and work up the options for the first workflow.
From you
Nothing.
A live session with your partners or leadership team on the draft findings. We argue it out, test assumptions, and settle the priorities together.
From you
Half a day from the sponsor and whoever decides.
The written report, the 90-day plan and the 12-month direction, presented live.
From you
One hour.
We agree what's in scope before day one and work from representative or de-identified matters wherever that's appropriate. We don't need your client files to understand your workflow; we need the people who run it.
A fixed fee, agreed before we start.
No hourly surprises. Scope is two to three workflows; we confirm the fee on the first call, and if your situation needs more, we'll say so and quote it before we start, not after.
The four days, the interviews with your process owners, the written report, the live walkthrough, and that every recommendation comes with its assumptions written next to it. If the report doesn't give your leadership team a decision they can defend, we'll rework it at our cost.
The productivity gain, the cost saving, or that your team will adopt the result. Those depend on decisions and follow-through inside your practice. Anyone who promises them is selling you the next stalled project.
You'll see an opinion in the report, so here it is up front. Keep the knowledge and the process in a form your practice owns and can take anywhere. Keep the model swappable. Frontier models trade places every few months; the way your practice prepares a trust distribution or a statement of advice doesn't. So: modular where it's cheap, and where it isn't, the cost of changing written down next to the recommendation.
That isn't a promise that switching is free. It's a promise you'll know the price before you commit — which is also what lets a partner defend the decision later, whichever way the market moves.
The review is a complete engagement on its own. Where it points — a company agent in the tools your team already uses, a knowledge layer your practice owns, a workflow with proper sign-off, or simply a cleaner process — you can build it with us, with your own team, or take the report to market and compare providers. The report is written so that any of the three works, and it says where further technical design is needed before anyone can quote properly.
What building with us looks likeNo — and the review will show you how to avoid having to. We start from the work and the systems you already have. Platform and model choices follow the workflow and the controls it needs, and the report says what each choice would cost to undo.
For a single task, often yes, and for some businesses that's enough. Use it. What AI can't do on its own is decide which work is worth pointing it at, reach your data safely with the right people seeing the right things, or tell you when its answer is wrong. That's the plan the review gives you.
Someone internal should own it. Building it is a different job: working out what AI can do for your business takes dozens of hours of trial and error, and on top of a full-time role those hours lose to client work every week. We put in those hours instead. We've already tested most of the paths that look right and aren't, starting on our own business, so you skip the U-turns. They own it; we keep it improving alongside them.
No. It identifies the operational and information-control questions the work raises, in language your compliance adviser or practice regulator can act on. They stay involved wherever their approval is required.
That's a good outcome at this price. The plan will say what to fix first — usually process, data ownership or access control — and you'll have avoided funding the next stillborn project.
Tell us where the time goes, what isn't connecting, or which automation stalled. Twenty minutes will tell us both whether the review is the right next step.
Existing client? Need remote support now?
Install Quick Connect, let us know, and we'll connect in seconds. Uninstall it once we're done.