Almost every accountancy practice knows AI is going to matter. They've probably had a play with ChatGPT; but they have no firm idea on how to implement it properly. That gap between awareness and action is exactly where most small and mid-sized UK practices are sitting right now.
The problem isn't interest or intelligence. It's that almost everything written about AI adoption is aimed at corporates with IT departments, innovation budgets, and a tolerance for jargon. For a practice of 5 to 50 staff, juggling self-assessment deadlines, FRS 102 accounts, and a team that already has too much on, "transformation initiatives" aren't the priority. Doing the work is.
This post lays out a practical starting framework. A way of thinking about where you are and what a sensible first step looks like. It won't take long, and it doesn't require any tools you don't already have.
Start with an honest assessment
Before you look at any AI tools, it's worth taking stock of where your practice actually stands. Most adoption failures happen not because the technology is wrong, but because firms underestimate how much the baseline matters. Four areas are worth considering honestly.
First, your technology stack. Are you cloud-based — using software like Xero, QuickBooks, or a cloud practice management platform — or are you still running desktop applications and shared network drives? This isn't a judgement; plenty of profitable practices run on desktop software. But cloud-based infrastructure tends to make AI integration more straightforward, and it's useful to know where you are before you start. Second, staff confidence. Have your team members actually tried any AI tools, even informally? A practice where one or two people have experimented personally is in a very different position to one where nobody has. Third, data hygiene. Is your client data reasonably well-organised, with clear naming conventions and consistent storage? Or is it scattered across inboxes, desktops, and shared folders with inconsistent naming? The messier the data environment, the more important it is to tidy up before introducing AI into workflows. Fourth, leadership appetite. Is a partner or director actively driving this, or is interest coming from more junior staff? Neither is wrong, but the path forward differs considerably.
A structured self-assessment beats guessing. Runbook's free AI Readiness Score works through exactly these questions in about two minutes and gives you a clear picture of where to focus, which is a more useful starting point than jumping straight to tool selection.
Pick one quick win, not ten
The most common mistake practices make when starting with AI is trying to do too much at once. A partner reads an article, gets enthused, and announces that the firm is going to use AI for client communications, tax research, meeting notes, marketing, and compliance review. Starting Monday. Nothing happens, because nothing specific enough has been defined to actually do.
A far more effective approach is to pick one repetitive task and try using AI for that single task for one week. Good candidates include drafting routine client emails (chasing missing information, confirming receipt of documents, acknowledging queries), summarising meeting notes from audio recordings, or drafting initial responses to straightforward HMRC correspondence. The task should be one that your team does frequently, that currently takes more time than it ought to, and where the output doesn't go directly to a client without any human review.
Starting small does two things. It builds genuine confidence. You develop a feel for where the tools are useful and where they fall short, based on real evidence rather than theory. And it creates internal proof of concept. The sceptical partner who doesn't believe AI will be any use is much more receptive to a colleague saying "I've been using it for client email drafts this week and it's saving me about 40 minutes a day" than to a promise about future capabilities. One important note: prompts need to be specific and UK-contextualised to produce useful output. A generic query like "write an email to a client about their accounts" will produce generic results. A prompt that explains the context, the type of client, the practice's tone, the specific situation, and any relevant UK-specific references, will produce something actually usable.
The data protection question
For most UK accountancy practices, the single biggest blocker to AI adoption isn't technology. It's data protection. And understandably so. You hold sensitive client financial information, and the thought of that data ending up somewhere it shouldn't is, rightly, a serious concern. The question, though, is more nuanced than "can we use AI at all?" The answer to that question is almost certainly yes in some form. The real question is which tools, for which tasks, with which data.
Using a free consumer-tier AI tool to process identifiable client financial data is a very different proposition from using an enterprise deployment where data processing agreements are in place and data isn't used for model training. Internal drafting tasks — summarising notes, generating first drafts from anonymised information, researching general questions — sit in a different risk category from tasks that involve feeding client-specific data directly into a tool. This isn't an area where a blog post can tell you what to do; that decision requires proper data protection advice specific to your practice's situation. What Runbook's AI Implementation Checklist does include is a structured questions framework to take into that adviser conversation, so you're asking the right things rather than starting from scratch.
A structured start beats a flying leap
AI readiness for an accountancy practice isn't primarily a technical challenge. It's an organisational one. The practices that will get the most from these tools over the next few years aren't the ones that move fastest, but the ones that move most deliberately: knowing where they stand, starting with something specific, and building from evidence rather than enthusiasm. If you haven't already, the free AI Readiness Score is the best place to start. It takes two minutes and gives you a clear picture of where to focus. For practices ready to go further, the AI Implementation Checklist (£97) provides the complete system: 10 sections, 14 checklists, 11 templates, and a 90-day plan built specifically for UK accountancy practices.