How to Use AI to Streamline New Client Onboarding in Your Practice

New client onboarding is one of the most document-heavy, time-consuming processes in any accountancy practice. A single new instruction can require a fee proposal, an engagement letter, a professional clearance request, an AML check, a software setup, and a welcome communication, and all of that happens before any billable work begins.

This article is part of Runbook's complete guide to AI tools for UK accountants. It covers exactly where AI can reduce the administrative load on your onboarding process, which tasks it handles well, which it does not, and the specific prompts your team can use right now. If you want to find out whether your practice is set up to use AI effectively more broadly, the free AI Readiness Scorecard gives you a personalised picture in under five minutes.

Why onboarding takes longer than it should

Ask any practice manager where unbillable time goes, and new client onboarding will feature prominently. The problem is not that the individual tasks are difficult. It is that there are many of them, they happen in sequence, and most involve producing written documents from a blank page each time.

A typical onboarding sequence for a new limited company client might include: an initial discovery call, a written fee proposal, an engagement letter tailored to the services agreed, a professional clearance letter to the outgoing accountant, AML and KYC checks, setup of the client record in practice management software, a welcome email with portal access instructions, and an internal handover note for the fee-earner who will manage the relationship. That is eight distinct tasks, most of them requiring someone to write something.

In practices without a systematic approach, much of this writing happens from scratch each time, with staff adapting old documents or piecing together templates they can find. The result is inconsistency, wasted time, and an onboarding experience that varies depending on who handles it rather than a process the firm can rely on.

AI does not remove the need for professional judgement in any of these steps. But it does remove the blank page problem entirely, and for a process as repetitive and structured as onboarding, that makes a material difference.

Where AI genuinely helps in onboarding

The tasks in onboarding that lend themselves best to AI assistance share a common characteristic: they follow a predictable structure, require clear written output, and benefit from having a professional review the draft rather than produce it from scratch. That describes the majority of the onboarding process.

Fee proposals

A fee proposal for a new prospect covers the same ground every time: a personalised opening, a clear breakdown of services and fees, what is included and what attracts additional charges, and a call to action. The structure is predictable; the personalisation is not. AI handles this well when you give it the specific details: business type, turnover, pain points mentioned in the discovery call, and services being quoted. The output is often a professional first draft in minutes rather than something built from a blank page.

Engagement letters

Engagement letters are one of the highest-value applications of AI in the onboarding process. They have a defined structure and legal significance, which means the stakes of getting them wrong are real, but the structure itself is consistent enough for AI to produce a strong first draft quickly. Our guide to using AI to draft engagement letters covers the process in detail, including worked prompts and a review checklist to use before any letter goes out.

Professional clearance letters

Professional enquiry letters usually follow a well-established format, informed by ICAEW guidance on change of professional appointment. The letter needs to confirm the client's instruction, ask whether there is anything of which you should be aware before accepting the appointment, and request the transfer of specific records. This is another task where AI produces a reliable first draft, and where the professional judgement required is in reviewing the draft rather than composing it.

Welcome communications

Once the engagement letter has been signed, the welcome email sets the tone for the client relationship. It needs to cover portal access, who the client's day-to-day contact will be, what information you need from them to get started, and what happens next. AI can usually produce a clear, professional welcome email quickly, ready for review and tailoring.

Internal SOPs and checklists

If your practice does not have a written standard operating procedure for onboarding, AI can produce a solid first draft of one in a single prompt. Prompt it with your practice size, services offered, software used, and AML procedures, and it will produce a numbered, stage-by-stage procedure document that your team can work from immediately and refine over time.

Prompts you can use right now

The following prompts are drawn from the AI Prompt Pack for UK Accountants, which contains 50 ready-to-use prompts across eight practice categories. Each prompt uses bracketed fields that you replace with the specific details for each new client instruction. Review all AI output before use. The prompts are a starting point, not a finished product.

Fee proposal (Prompt 02)

This prompt asks you to supply the prospect's business name and type, approximate annual turnover, number of employees, pain points raised in the discovery call, and the services being quoted. The output is a one-page fee proposal covering a personalised opening, itemised services and fees, a value proposition, and a 30-day validity period. It consistently produces a professional first draft that requires editing rather than composing.

New client engagement letter (Prompt 01)

This prompt requires the client name, entity type, services engaged, fee, and applicable accounting standard. The output covers scope of services and exclusions, client responsibilities, your responsibilities, fees and payment terms, complaints procedure, ICAEW regulatory status, a data protection statement, and a signature block. It is structured as a formal engagement letter in plain British English. Always review against your firm's approved engagement terms before issue.

Professional clearance letter (Prompt 28)

This prompt generates a professional enquiry letter to the client's previous accountant, covering the confirmation of instruction, the request for any information you should be aware of, and a tailored list of records to be transferred. You specify the records relevant to the services you are taking over: last set of accounts, trial balance, depreciation schedules, VAT and PAYE references, and any outstanding tax correspondence. The output is appropriate for inter-professional communication and requires a senior review before sending.

Client onboarding SOP (Prompt 27)

This prompt produces a full standard operating procedure for your onboarding process, structured by stage: initial enquiry handling, fee proposal, AML and KYC checks, engagement letter, software setup, handover from previous accountant, welcome communication, and internal sign-off. Each stage includes responsible person and estimated time fields. Supply your practice size, services, software used, and AML approach and the output is a procedure document your team can use as a working draft immediately.

What AI cannot replace in onboarding

There are parts of the onboarding process where professional judgement is irreplaceable and AI should not be in the loop at all.

AML and KYC checks

Anti-money laundering and know your client checks are a compliance obligation under the Money Laundering Regulations 2017. They involve electronic identity verification, sanctions screening, PEP checks, and a client-level risk assessment. General-purpose AI tools should not be used to carry out, verify, or evidence these checks. Use your approved AML software and documented firm procedures, with appropriate professional review. AI may help draft internal process notes or onboarding checklists, but it should not make or validate the AML decision. The AI Implementation Checklist covers the compliance considerations for AI use in practice in detail, including what must be kept entirely outside AI workflows.

Accepting or declining an instruction

The decision to accept a new client requires professional judgement: assessing the client's business, considering any ethical conflicts, reviewing the information received from the previous accountant, and making a considered decision to proceed. That judgement belongs to a qualified professional. AI can draft the documents that follow the decision, but it cannot make the decision.

Handling information from the previous accountant

When the outgoing accountant responds to your professional clearance request, what they say, or do not say, requires careful professional assessment. If information is disclosed that gives you cause for concern, that is a matter for a qualified person to consider, not an AI tool to summarise.

Reviewing engagement letter terms

An AI-generated engagement letter is a first draft. Before it goes to any client, a responsible person at your practice must review it against your approved engagement terms, confirm it accurately reflects the scope agreed, and check that the data protection, liability, and regulatory wording meets your practice's standards. The output is a drafting aid, not a finished legal document.

Data protection: what to check first

Onboarding involves a higher concentration of personal data than most other practice workflows. You are collecting the client's details, their business information, and potentially information about their financial position and history, and all of this arrives before any services have been delivered. This makes data protection a non-negotiable consideration before introducing AI into the process.

Free-tier AI tools are unlikely to provide the contractual protections, data processing terms, administrative controls, and assurance most accountancy practices would need before processing identifiable client data. Treat them as unsuitable for real client data unless your practice has specifically reviewed the tool, its terms, and its UK GDPR implications.

Do not assume that any paid AI subscription is suitable for client data. The key question is whether the tool is being used under business or enterprise terms with appropriate data processing terms, security controls, retention settings, and administrative oversight. Examples may include ChatGPT Business or Enterprise, Microsoft 365 Copilot or Microsoft 365 Copilot Chat used through an eligible Microsoft 365 commercial account, and Claude for Work under Anthropic's commercial terms. Even then, your practice should review the specific terms and configuration before processing identifiable client information.

The safest approach during the initial stages of AI adoption is to use anonymised or placeholder data in all AI-assisted drafting. Produce the draft with placeholder fields, then populate it with real client details only within your secure practice management system. This approach is entirely workable and avoids the compliance risk entirely while you put the right infrastructure in place.

For advice specific to your practice's obligations, consult a qualified data protection adviser. Runbook does not provide GDPR or legal advice.

Building a repeatable AI-assisted onboarding workflow

The practices that get the most from AI in onboarding are not the ones that use it occasionally on an ad hoc basis. They are the ones that have built a defined workflow: a documented sequence of steps, a set of agreed prompts for each document type, a clear review process, and a consistent standard for what leaves the practice.

A practical starting point is to choose the single most time-consuming document in your current onboarding process and introduce AI drafting for that task alone. For most practices, that will be the fee proposal or the engagement letter. Use it for the next five new client instructions, refine the prompt based on what comes back, and establish a review standard. Once that is working consistently, extend the approach to the next document in the sequence.

The goal is a process where any member of staff can open a prompt, supply the relevant client details, produce a professional first draft, review it against a checklist, and have a document ready for partner sign-off in a fraction of the time it currently takes. That is not a distant ambition. For the document types described in this article, it is achievable within weeks of starting.

A structured approach to implementing AI across your practice, including onboarding and all other key workflows, is set out in the AI Implementation Checklist for UK Accountancy Practices. It covers tool selection, data protection preparation, staff rollout, and a 90-day implementation roadmap. If you are not yet sure where your practice stands, the free AI Readiness Scorecard identifies your starting point and what to prioritise first.

Frequently asked questions