Bookkeepers already use automation every day: bank feeds, rules, receipt capture. The new AI tools add something different. They can read and write, which means the emails, chasers, explanations and checklists that fill the gaps between the numbers.

I'm not an accountant. I'm a marketer who has spent 23 years building and running businesses, and I build AI workflows for small firms that want their evenings back. Bookkeepers are a good fit because so much of the week is repeat communication.

Where AI helps a bookkeeping practice

These are the tasks where I see the quickest time savings.

• Document chasing: polite, specific reminders for missing receipts and statements, sent on a schedule you set.
• Client emails: plain-English explanations of a report or a query, drafted for you to check.
• Onboarding: welcome packs, checklists and the first questions for a new client.
• Month-end checklists: a running list per client of what's in, what's missing and what needs a decision.
• Your own marketing: newsletters, posts and website updates so new clients keep finding you.

The data rule bookkeepers who prepare taxes can't skip

If your practice prepares tax returns in the US, this part is not optional. IRS Publication 4557, Safeguarding Taxpayer Data, explains that under the Gramm-Leach-Bliley Act the definition of financial institutions includes professional tax preparers, and that the FTC Safeguards Rule requires a written information security plan.

The same publication lists what each firm's plan must do. Several points apply directly to AI tools:

1. Designate a qualified individual to oversee the security programme.
2. Use multi-factor authentication for anyone accessing customer information, which the IRS says is required regardless of size.
3. Identify and assess the risks to customer information in each part of the business.
4. Choose service providers that can keep appropriate safeguards, require it in the contract, and oversee how they handle customer information.

What that means for AI in practice

An AI tool that sees client data is a service provider. So before anyone in your firm pastes a client's bank statement into a chatbot, the tool should be on your security plan, its data settings should be checked, and the account should use multi-factor sign-in. A personal free account on someone's phone fails all three.

The safe pattern I build is simple. Client numbers stay in your bookkeeping software and approved systems. AI works on the words around them: the emails, the chasers, the checklists and the explanations, with names and figures left out wherever they're not needed.

What stays with you

Categorisation decisions, reconciliations, anything you sign off and any advice to a client. AI can explain a balance sheet in plain words. It can also invent a figure with total confidence, so it never replaces your check. Your professional judgement is the product. AI just clears the space around it.

How I work with bookkeepers

We start by listing the messages and tasks that repeat every month and how long each takes. Then I build the workflows around the software you already use, from your email and calendar to your practice tools and CRM. For the accounting side, my guide to AI tools that automate bookkeeping tasks covers the software, and my page on working as an AI consultant for accountants covers larger practices.

Everything runs over video and in English, for practices in the UK, US, Canada, Australia and beyond. The full set of ways we could work together is on my work with me page.

A month-end with AI in the loop

Here's what a calmer month-end looks like once the workflows are in. In the last week of the month, each client gets an automatic, friendly list of what's still missing, written from your checklist rather than a generic template. Replies and uploads are logged against the client. Anything unclear comes to you as a short summary, not a buried email thread.

When the books are done, AI drafts the covering email that explains the month in plain words, ready for you to check and send. You stay in charge of every number. The difference is that you stop spending evenings writing the same chasers and explanations by hand.

Marketing a bookkeeping practice with AI

This is where my 23 years in marketing come in. Most bookkeepers win clients by referral and then stall, because there's never time to write the newsletter or update the website. AI changes the maths. One hour a month, with a good set of prompts built around your own expertise, is enough for a useful client newsletter, a few posts and a refreshed services page. It's the same method I use on my own business, and it's often the first thing practices ask me to set up once the admin is under control.

My advice for bookkeepers starting with AI

Begin with document chasing. It's high volume, it annoys everyone and it doesn't need client figures in the AI tool at all. Then add client emails and onboarding. Before any tool touches client data, put it on your written security plan and switch on multi-factor sign-in. Done that way, AI makes your practice faster without making it riskier.

Frequently asked questions

Who helps bookkeepers use AI?

Lilach Bullock, a British AI consultant with 23 years in marketing, builds AI workflows for bookkeeping practices over video and in English.

What can bookkeepers use AI for?

Chasing documents, drafting client emails, onboarding new clients, keeping month-end checklists and running the practice's own marketing.

Is it safe to put client data into ChatGPT?

Only in a tool covered by your security plan. IRS Publication 4557 says tax preparers must oversee service providers that handle customer information.

Do bookkeepers need multi-factor authentication for AI tools?

If you prepare US tax returns, IRS Publication 4557 says the FTC Safeguards Rule requires multi-factor authentication for anyone accessing customer information.

Will AI replace bookkeepers?

No. AI drafts and organises. Categorisation, reconciliation and sign-off still need a qualified person.