We Don’t Know Where to Start with AI:
What Should We Do First?

Your organisation knows it needs to do something about AI but nobody can say exactly what. Most leadership teams arrive at this point the same way: pressure from the board or partners to act, an abundance of tools and opinions, and no obvious first step. This page explains why the uncertainty happens, what a sensible starting position looks like, and how to work out where to start with AI in your firm, in plain terms, with free resources you can use today.

Why firms don’t know where to start with AI

We see the same uncertainty in well-run firms across every sector. It comes from where this technology currently sits, and it has five recognisable causes.

• There is more advice than clarity. The volume of tools, opinions and offers is enormous, and very little of it is organised around your situation. Working out what applies to a firm like yours is a genuine analytical task, and most leaders simply have not had the time to do it.
• Most guidance assumes an enterprise. Published case studies tend to involve in-house data teams, dedicated change budgets and a CTO with capacity to run pilots. For a 40-person practice, or a 200-person business where IT is one capable person and a support contract, the playbook has to look different, and very little has been written about what that version looks like.
• Ownership is hard to assign. AI does not sit neatly in anyone’s existing role. Enthusiasm exists in most firms, but enthusiasm without dedicated time and decision-making authority rarely turns into a plan. This gap is structural, and in our experience it is the single most common reason progress stalls.
• The risk questions are legitimate. Client confidentiality, professional indemnity, regulator expectations: these deserve proper answers, and they are rarely available in one place. Until they are answered, caution is the responsible position, even though it can feel like inertia.
• Every option looks like a commitment. Pilots, platforms, subscriptions, retained support. Without a way to compare them, it is hard to tell a good first step from an expensive detour, so decisions wait. Meanwhile, staff are often already using free AI tools informally, which means the firm has exposure without the benefit of a considered position.

If you have already made a start

Most firms reading this page have already done something, and that is worth saying plainly: nothing you have done so far is wasted. It usually just needs connecting to a plan.

• If you have Copilot or similar licences, you have made a sensible early investment. The value arrives when there is a clear view of where those tools save time, who should be using them for what, and how usage is supported and measured. With that in place, existing licences often become the quickest first win available.
• If a partner or team member has been experimenting, you have something genuinely valuable: first-hand insight into where the friction sits in your own work. The next step is giving that insight structure, so it can spread beyond one person and survive their busy periods.
• If you have commissioned external thinking, whether a strategy piece, a readiness review or specialist advice, you have a foundation. The most common gap is simply the bridge from recommendation to delivery, and that bridge is buildable.

Whatever the starting point, the task is the same: take stock of what exists, decide what it is for, and sequence what comes next.

What a clear AI starting position looks like

A firm with a clear starting position knows three things. First, where it actually stands today, including the tools already in use, formally and informally. Second, the two or three things genuinely worth doing first, chosen because they free up fee-earner or specialist time, or remove a real bottleneck. Third, the things it has deliberately decided to leave alone for now, with reasons written down. Every cost is understood in pounds and weeks. Risk has been considered against the standards your regulator and insurer would expect. And when AI comes up at the next partner or board meeting, there is a plain answer ready. That position is achievable in about a month, and it does not require a strategy document, a transformation budget, or a hire that is difficult to make.

How to work out where to start with AI: the first 30 days

1. Write down the question you were actually asked. “What are we doing about AI?” is usually code for something more specific: a client asked, a competitor said something visible, a colleague pushed, or the board read something striking. Pin down what triggered the pressure, because that defines what a credible answer needs to contain. An answer to the wrong question, however polished, will not hold the room.
2. Take an honest baseline, including the informal picture. List what is already in use: the practice or operations systems, the document tools, any Copilot or equivalent licences, and the personal AI habits that rarely come up in meetings. Most firms discover more AI in the building than they expected. Treat that discovery as useful data: it shows you exactly where your people already feel the friction, which is the best guide to where help is wanted.
3. Map where the time goes. Start with the work your firm already does. List the processes that consume the most professional or administrative hours: client onboarding and compliance checks, engagement letters, first drafts of routine correspondence, meeting notes, timesheet narratives, document review, candidate screening, and reporting. Pick the two or three that are high volume, low judgment, and well documented. That is where a first project earns its keep, and where time savings are easiest to evidence to a careful partnership or board.
4. Answer the governance question properly, once. Before anything touches client or customer data, establish four things: which tools offer business terms with no training on your data, where data is stored and for how long, what your regulator has said about AI use (the SRA, ICAEW and FCA have all published positions, and most sector regulators now have), and whether your professional indemnity insurer would like to be informed. This takes days, not months, and once done it converts a vague sense of risk into a short, workable list of conditions any tool must meet.
5. Rule things out deliberately. Ruling things out is a decision in its own right, and it belongs in writing. Client-facing advice generation, anything touching privileged material without human review, building bespoke models: for most firms, these sensibly wait beyond year one. Write the list down with one line of reasoning per item. It is half of a credible plan, it is the half that reassures a careful partnership, and it is the half most firms forget to write down.
6. Get one senior, independent view before committing further. An hour with someone who has seen this across many firms, and who is not attached to any single tool or platform, gives you a reference point for everything above: the baseline, the first steps, the leave-alone list. Firms that take this step typically reach a workable plan in weeks rather than quarters.

Taking AI to your partners or board

Often, bringing your organisation, partnership or board on the AI journey can be the biggest challenge to overcome. The format that works is short and concrete: one page covering the honest baseline, two or three proposed first steps with costs in pounds and timescales in weeks, the leave-alone list with reasons, and the governance position in plain language. Boards have seen open-ended technology commitments before, and they are right to be wary of another one. Give them something bound, costed and reversible, and the conversation changes character. The editable slides below are built for exactly that meeting.

Where to start with AI: the questions leaders ask

Do we need an AI strategy before we do anything?

Not at the outset. For most organisations of this size, what is needed first is a baseline and two or three well-chosen first steps. Strategy then emerges from doing something small properly and building on what the firm learns. A strategy written later, informed by real experience, is worth far more than one written before the firm has learned anything.

We have already bought Copilot licences. Does that count as started?

It is a start, and a common one. The step that unlocks the value is a plan: which tasks the licences should be saving time on, who should be using them, and how that usage is supported. Many firms find their existing licences become the quickest first win once that plan exists, which makes them a good investment that simply arrived ahead of its instructions.

How much should a firm our size expect to spend on a first step?

Less than most leaders fear. A sensible first phase, meaning a baseline, a plan and one working improvement, is typically a five-figure commitment measured in weeks rather than a six-figure programme measured in quarters. Larger commitments make sense later, once there is evidence from your own firm to justify them.

Is it safe to use AI with client data?

It can be, with the right setup, and the setup matters because informal use of free tools is often already happening. The useful question is how to replace informal use with a configuration your regulator and insurer would recognise as competent. That is achievable, and usually quicker than firms expect.

Who should own AI internally?

Someone senior enough to make decisions and protected enough to have the time. In most organisations, that means the COO, practice manager or operations lead, with explicit partner or board backing. Ownership means a named person, a small budget and a standing agenda item. With those three in place, even modest enthusiasm compounds.

What does a first conversation actually involve, The Recce?

At Ai Terrain it is called The Recce: one hour, free, with a senior advisor, your business on the table, no slides and no pitch. Within 48 hours you receive a plain note covering where you stand, what is worth doing first, what to leave alone, and what a sensible next step would cost. It is useful even if you never speak to us again.

The fastest way to a clear starting position is a conversation.

Request a Recce: One hour, free, with a senior advisor and your business on the table. You will leave with an honest view of where you stand, what is worth doing first, and what a sensible next step would cost. No slides, no pitch.