The KPMG, Baker Tilly and Fieldguide AI-Native Firm Advantage Report is more revealing for what it doesn't say than what it does. Yes, firms with no AI plans fell from 28% to 19% in a year, and active users grew from 22% to 40%. That's real movement. But it tells us nothing about whether those 40% are using AI responsibly under FCA Consumer Duty PS22/9, SRA Code of Conduct, or the FRC's emerging ISA UK requirements for audit. For mid-market UK regulated firms, this is the question that matters: are you adopting AI, or are you adopting liability?

What this data reveals is a profession in the middle of a forced migration, not a considered transition. The 9-point swing away from 'no plans' in a single year suggests firms are making adoption decisions because competitors are moving, not because they have solved the governance problem first. This is exactly what happened with cloud adoption in the mid-2010s—firms rushed in, then regulators had to wade in behind them. The EU AI Act is live. The ICO has issued guidance on UK GDPR and AI. The PRA's SS1/23 direction on third-party risk now explicitly covers AI vendors. Firms jumping without thinking about governance are not ahead; they are exposed.

Here is Trovix's honest take: most generative AI tools sold to accountancy firms today—whether it is Microsoft Copilot, Harvey, Luminance, or generic LLM wrappers—are being deployed as productivity widgets. They are plugged into intake, document review, or draft generation without a single framework for control, audit trail, or quality gates. They work until they don't. And when they don't—when they hallucinate a tax code, miss a compliance flag, or serve up advice that breaches SRA protocol—the firm has no governance layer to explain what happened or why. Trovix Audit exists precisely because this gap is fatal for regulated firms. You cannot adopt AI at scale without knowing what the AI did, why it did it, and who is liable when it fails.

What a mid-market accountancy or financial services firm should do right now is stop asking 'which AI tool is best' and start asking 'how do we implement any AI tool safely'. That means: establish an AI governance framework before you deploy anything new; map your current AI use (including ChatGPT and Copilot in the wild); assess each tool against FCA Consumer Duty, SRA Code, and ICO guidance; build audit and explainability into your workflow; and use Trovix Watch to monitor regulatory changes as they land—because they will land faster than your deployment cycle. The firms winning the adoption race are not the ones with the most tools. They are the ones who treated AI governance as a day-one problem, not a day-90 remediation.

Source: Accounting Today