Can AI write a client engagement letter?

AI can produce a strong first draft if you supply the engagement specifics. It cannot replace the partner-level review that adjusts liability language, scope boundaries, and indemnification clauses. Treat AI output as a starting point and never send a client engagement letter without a senior-level review.

Is it ethical to use AI to draft an audit findings memo?

Generally yes, with caveats. The findings, conclusions, and judgments must be yours — AI is drafting the prose, not making the call. Most firms now require AI-assisted documents to be reviewed end-to-end by a human and the AI involvement noted in the workpaper.

AI for Client Communication: Drafting Engagement Letters, Audit Memos, and Follow-Ups

Client-facing writing is one of the most under-appreciated time sinks in accounting work. Engagement letters, audit findings memos, tax-position letters, follow-up emails, year-end management letters — they consume hours that aren't recoverable on the bill. AI is finally a credible time-saver here, if you use it right.

The three modes of AI-assisted writing

Draft mode: AI writes the first version; you edit heavily.
Polish mode: You write the first version; AI tightens it.
Translation mode: AI converts technical accounting language into client-readable plain English (or vice versa).

Most accountants overuse draft mode and underuse polish and translation modes. Polish mode produces better client communication because the substance comes from the human and the AI just refines tone.

The "client memo" prompt template

"You are drafting a memo for [client industry] from [firm name]. Tone: professional, concise, neutral. Audience: CFO who is not an accountant by training. Topic: [topic]. Key points to cover, in order: [bullet list]. Length: 3-4 short paragraphs. Don't include disclaimer language — that's added separately. Don't use the word 'leverage.'"

The constraint on filler words is critical. AI drafts by default include "leverage," "synergize," "robust" — words that signal AI authorship to senior partners.

The engagement letter trap

Engagement letters are legal documents. AI is fine for the descriptive sections (scope, timeline, responsibilities). AI is dangerous for liability, indemnification, and limitation-of-damages clauses — those are where firms have specific approved language that's been reviewed by counsel. Pulling those from an AI risks substituting unreviewed terms.

Best practice: AI drafts scope and timeline; firm boilerplate handles liability sections.

Tone and voice consistency

AI-drafted communication often loses the firm's voice. Before going wide with AI assistance, give it three or four examples of how your firm writes to clients. Tell it to "match this voice" in subsequent drafts. Voice consistency matters for client trust.

The "polish mode" pattern

"I've drafted the memo below. Tighten it: shorter sentences, remove hedging, no cliches, keep all factual content. Length should drop by 20-30%. Don't add new claims."

This pattern is almost always better than asking AI to draft from scratch — the substance stays yours.

What never to do

Send unedited AI prose to a client. Clients can tell, and it damages trust.
Paste client confidential data into a public AI tool. Use your firm's sanctioned AI environment.
Use AI to write tax-position language without a partner review. Liability-bearing language needs human judgment.
Skip the disclosure step if your firm requires noting AI assistance.

The compounding skill

Accountants who learn to drive AI for client communication well save 3-5 hours per week and produce better-written deliverables than they would have without AI. That's a 10-15% productivity bump that compounds over a career — and most of your peers are still typing every memo from scratch.