

AI governance for professional services firms, without enterprise bureaucracy.
Practical rules your team will follow covering who owns AI, which tools are approved, how client data is handled and who reviews outputs. Start with our free template.
4.9+ from 226 reviews
What an AI governance consultant does for a 10-200 person firm.
AI governance sets out who can use AI, for which tasks, with what data, and who checks the output. For a firm of 10 to 200 people it fits on a few pages. As your AI governance consultant, we draft those rules with you, configure your tools to enforce them and train your team so day-to-day practice matches the policy.
Ownership
One named partner owns AI, with a deputy. They approve new tools and uses, maintain the register and report to the partners.
Approved tools
A short list of business-tier accounts that don't train on your data, keeping client work off personal ChatGPT logins.
Data rules
Clear categories for public, internal, client-confidential, privileged and special category data, matched to the tools permitted to process them.
Human review
A qualified team member reviews AI drafts before they reach a client, a court, HMRC or the opposing side.
Audit trail
A record of what's approved and why, plus where client drafts came from and who signed them off.
Training
Staff learn the rules and how to verify AI output before getting access, with refreshers when tools change.
Incident handling
A clear process if client data enters an unapproved tool or an error reaches a client, setting out who to notify and how quickly.
Most 10-200 person firms can skip AI ethics boards, lengthy policy manuals and voluntary certifications built for large enterprises with dedicated AI teams. Putting the seven controls above into practice covers what regulators and clients expect. To see which of them your firm already has, score yourself against our AI readiness checklist.
Free AI governance framework template. Copy it and adapt it to your firm.
Ten sections you can copy into Word and complete, each referencing the underlying framework. Treat it as a practical starting point rather than legal advice, and have your COLP, compliance partner or DPO adapt and own the final version.
- 01
Scope
Draws on: UK AI regulation principles: accountability and governance
- Who this covers: partners, staff, trainees and contractors.
- What counts as AI: chat assistants, AI inside Microsoft 365 or Google Workspace, AI features in your practice or accounting software, and custom workflows built for you.
- The date the partners approved it, and the date of the next review.
- 02
Ownership
Draws on: NIST AI RMF: Govern. ISO/IEC 42001: roles and responsibilities
- The partner who owns AI, and their deputy.
- Who approves a new tool, and who approves a new use of an approved tool.
- How often the owner reports to the partners, and reviews this policy (at least twice a year, and whenever a tool or rule changes).
- The firm remains responsible for all client communications and advice, regardless of which tools assisted with drafting.
- 03
Approved tools register
Draws on: ISO/IEC 42001: AI system inventory. NIST AI RMF: Map
- For each tool: what it's for, business or personal account, who owns it, where the data is stored, whether the provider trains on your data, and the date it was approved.
- Client work only goes into approved tools on business accounts.
- How to ask for a new tool, and who answers.
- 04
Data rules
Draws on: UK GDPR. ICO guidance on AI and data protection
- Your classes of data: public, internal, client confidential, and special category or privileged.
- Which class can go into which approved tool.
- A data protection impact assessment (DPIA) before AI processes client personal data.
- What each tool keeps, for how long, and in which country.
- 05
Risk assessment for each use
Draws on: NIST AI RMF: Map and Measure. UK AI regulation principles
- The task, who does it now, and the data it touches.
- What could go wrong: a wrong answer, a leak, unfair treatment, or people trusting it too much.
- How likely and how bad, the controls you'll use, and what risk is left.
- Who signed it off. The AI owner can approve low-risk internal uses, while client-facing advice or decisions affecting individuals go to the partners.
- 06
Human review
Draws on: SRA Risk Outlook on AI. ICAEW Code of Ethics: professional competence and due care
- The role that checks each kind of output, by name.
- What they check: facts, figures, citations, tone and confidentiality.
- AI-assisted drafts do not go to a client, court, regulator or opposing party until a qualified person has approved them.
- 07
Telling clients
Draws on: UK AI regulation principles: transparency, contestability and redress
- What you tell clients about AI, in your engagement letter or when they ask.
- How a client can raise a concern about AI on their work, and who answers.
- 08
Audit trail
Draws on: ISO/IEC 42001: documented information. ICO: accountability
- Keep the tools register, risk assessments, DPIAs and training records in one place.
- For client work, record where each AI draft came from and who approved it, on the matter or client file.
- 09
Training
Draws on: SRA Risk Outlook on AI. NIST AI RMF: Govern
- Staff complete training on the policy and how to check AI output before getting access.
- Refresh it when tools or rules change, and record who has done it.
- 10
Incidents
Draws on: UK GDPR Article 33. NIST AI RMF: Manage
- What counts: client data in an unapproved tool, an AI mistake that reached a client, or a provider changing its terms.
- Who to tell, and how fast. A personal data breach that risks people's rights must be reported to the ICO within 72 hours.
- Work out what happened, fix it and update these rules.
The frameworks behind the template, and how much weight each one carries.
UK GDPR and your regulator's rules are binding, while the standards below are voluntary frameworks you can draw on. None of them requires a 10-200 person firm to seek certification.
- UK AI regulation principlesFive principles from the government's March 2023 white paper: safety, security and robustness; transparency and explainability; fairness; accountability and governance; contestability and redress.Not law. Existing regulators, such as the ICO, apply them in their own areas.
- ICO guidance on AI and data protectionHow UK GDPR applies when AI uses personal data: accountability, lawful basis, fairness, security, DPIAs and people's rights.UK GDPR is law. The guidance is the ICO's view of how to comply, and is under review after the Data (Use and Access) Act 2025.
- NIST AI Risk Management FrameworkFour functions, Govern, Map, Measure and Manage, from the US standards body (January 2023), with a Generative AI Profile added in July 2024.Voluntary. A useful checklist in any country.
- ISO/IEC 42001:2023The international standard for an AI management system: policy, roles, risk and impact assessment, and continual improvement. Same structure as ISO 27001.Voluntary. You can be certified against it, but nothing requires a firm to be.
AI governance for law firms, accountants and consultancies, mapped to what your regulator expects.
Neither the SRA nor ICAEW bans AI. Both expect the core controls in the template: clear accountability, confidentiality safeguards, review by a qualified person, and staff trained to use and check the output.
Law firms
The SRA
The SRA allows AI use provided you stay responsible for what it produces and do not delegate that accountability to IT or a supplier. Its Risk Outlook report highlights protecting confidentiality and privilege, supervising AI and the staff using it, keeping training up to date, and telling clients when AI is used on their matter. The report itself is guidance alongside the binding SRA Standards and Regulations.
Accountants
ICAEW
ICAEW's Generative AI guide covers confidentiality, risk and governance. While the guide is advisory, the Code of Ethics applies regardless of the tool used, including confidentiality and professional competence and due care, which extend to training and supervising your team. A qualified person still signs off client work, and your MLRO retains responsibility for AML decisions.
Consultancies
UK GDPR and your contracts
Consulting has no UK sector regulator. UK GDPR, client contracts and NDAs set the boundaries, so the template's sections on data rules, approved tools and client wording matter most.
What we do as your AI governance consultant, as part of the AI we set up.
We include governance in the NQU AI Accelerator, writing your policy alongside the workflows we build so the rules reflect the tools and habits your team uses.
AI opportunity audit
We map the tools your team currently uses, how each handles your data, and where AI will save the most time.
Your AI policy
Rules, prompts and client wording, written to what your regulator expects, for your COLP or compliance partner to adapt and own.
Compliance support
Guidance on which tools suit each category of client data.
AI SOPs
A one-page guide for each Task Kit covering the approved tool, required inputs and expected output.
Human review built in
AI produces the first draft, and a member of your team approves it before it goes to a client.
Audit trail
Drafts record their source material and are filed against the relevant matter or client.
Data in your accounts
It runs in your own Microsoft or Google account, or in ours in the UK, EU or US. Business AI accounts only.
Training and helpdesk
Workshops on request, and answers in Teams or by email when your team is stuck.
We're not a law firm and this is not legal advice. Your COLP, compliance partner or DPO owns the policy, while we help draft it and configure your tools to match.
Frequently asked questions.
Have another question? Ask us on the intro call.
An AI governance consultant helps a firm decide who owns AI, which tools are approved, how client data is classified, who checks outputs and how incidents are handled. They then configure the tools to match and train the team. For a firm of 10 to 200 people, that usually comes down to a few practical pages of rules staff can follow.
Yes. The template on this page covers ten sections: scope, ownership, an approved tools register, data rules, per-use risk assessments, human review, telling clients, an audit trail, training and incidents. Copy it into Word and adapt it with your compliance lead; it is a practical starting point and does not constitute legal advice.
For a firm of 10 to 200 people, seven areas cover the essentials: a named owner, an approved tools register, data-handling rules, human review of client outputs, an audit trail, training and an incident plan. Larger standards such as ISO/IEC 42001 add formal objectives, internal audits and management reviews, which you can adopt later if needed.
One named partner with a deputy. The SRA makes clear that firms can't delegate accountability for AI to an IT team or external provider. In a law firm the COLP is usually involved, while in an accounting firm it's often the partner who owns compliance. They approve tools and use cases, keep the register and report to the partners.
Yes, if staff use AI on client work, including through personal ChatGPT accounts. A smaller firm only needs a concise policy, an approved tools register and a named owner. Start with data rules and approved accounts, as unapproved tools are where client information typically leaks.
Write one short assessment for each specific use case. Record the task, the data involved, what could go wrong, how likely and severe those risks are, your controls and who signed it off. Section five of the template gives you a ready-made format, and you should add a DPIA whenever personal data is processed.
Usually not. ISO/IEC 42001 is a voluntary standard for an AI management system, and nothing in UK law or in SRA or ICAEW guidance requires a firm to be certified. It's worth borrowing its structure. Certify only if your clients ask for it, much as some ask for ISO 27001.
Pricing depends on the size of your firm. We cover it on the intro call and confirm the exact figure in your AI opportunity audit. Governance is included in the NQU AI Accelerator, under the same monthly fee as the rest of our work.

See where your firm stands.
Paste your website to see how much of your firm's work AI can already handle, then score your policy, data and people. Ready to build? See the AI Operating System.