6 min read
This is an illustrative composite case study. Marisol is not a real person; her situation is built from common patterns in solo bookkeeping practices, and the numbers are realistic examples rather than reported results.
It is the second week of February, and Marisol is staring at a spreadsheet with 38 client names and a column she has started to dread: “Docs received?” Eleven say yes. The rest are a mix of “partial,” “promised Friday” and blank. Her CPA partners want clean books by early March. Her clients want to know why she keeps emailing them. And she is spending more time chasing paperwork than doing the bookkeeping she is actually paid for.
That was last year. This year she decided to fix the chase before it started. Here is what she did in late November, what AI handled, what she refused to let it touch, and how the January picture changed.
The real problem was not lazy clients
When Marisol looked back at her emails from the previous season, a pattern jumped out. Her clients were not ignoring her. They were confused. Every client got the same generic “please send your year-end documents” email, and every client had to work out for themselves what that meant for them. The landscaper with two part-time helpers needed different paperwork than the Etsy seller with one bank account, but both got the same message.
So the clients did what people do with vague requests: they put it off until they had time to figure it out. The fix was not more reminders. It was clearer, client-specific requests sent earlier.
Step 1: Build each client’s checklist from last year’s file list
Marisol already had the raw material. For every client, her folder from the previous year showed exactly what had eventually been sent: bank statements, credit card statements, loan statements, contractor W-9s, payroll reports, and so on.
She exported just the file names for each client (no contents, no account numbers) and pasted them into an AI assistant on a business plan whose settings keep her chats out of model training. Her prompt was simple:
“Here is the list of documents this client sent last year for year-end bookkeeping. Turn it into a plain-English checklist for this year, grouped by where the client will find each item (their bank’s website, their payroll app, their email). Add a line asking whether anything changed this year, such as a new account, a new loan or a new contractor. Keep it under 150 words and friendly.”
In about two evenings she had 38 tailored checklists. She read every one, fixed the handful where the AI had guessed wrong (one client had closed an account, another had switched payroll providers), and added a note to her own client file for each change.
Step 2: Front-load the request into December
Instead of asking in January, when everyone is buried, she sent the checklists in the first week of December with a clear message: “Nothing to send yet, this is just so you know what’s coming. The first items will be ready on your bank’s website in early January.”
She also flagged the deadline that tends to sneak up on small business owners: if a client paid an independent contractor during the year, Form 1099-NEC is generally due to both the contractor and the IRS by January 31, according to the IRS page for Form 1099-NEC. For clients who used contractors, she asked for any missing W-9s in December, when people actually answer emails, rather than in late January.
Step 3: Reminders that say what is still missing
Marisol tracked each client’s checklist in her spreadsheet. Every two weeks from early January, she asked AI to draft a short follow-up for each client with outstanding items, listing only what was still missing:
“Thanks for the bank statements! Two things left: your December credit card statement and the loan statement from your lender. Both should be under ‘Documents’ when you log in.”
She sent them herself, in batches, after a quick read. Specific beats generic: a reminder that names two items gets acted on; a reminder that says “please send your documents” gets snoozed.
What she refused to hand to AI
This is the part that kept the project safe. Bookkeepers handle exactly the kind of information that identity thieves want, and the IRS spells out expectations for protecting it in Publication 4557, Safeguarding Taxpayer Data. Marisol set three rules:
- No identifiers ever go into a chat: no Social Security numbers, EINs, account numbers or full statements. File names and plain descriptions only.
- Documents travel through her secure client portal, never as email attachments and never through the AI tool.
- Every client-facing message gets a human read before it goes out, because a wrong checklist item wastes a client’s time and her credibility.
She also told clients, in one sentence in the December email, that she used an AI assistant to help draft checklists and reminders and that none of their financial documents or numbers were shared with it. Nobody objected. Two clients asked how they could do the same thing for their own businesses.
The January picture
In this composite example, the change is the kind a solo bookkeeper could reasonably expect:
- By January 31, 29 of 38 clients were complete, compared with 11 by mid-February the year before.
- Her follow-up emails dropped by roughly half, because each one asked for something specific.
- Every 1099 client had W-9s on file before the deadline, instead of the usual last-week scramble.
- The time she spent on checklists in November (about six hours, including review) came back several times over in February.
The deeper win was the relationship. Clients who had felt nagged now felt guided. The checklist turned a dreaded chore into a list they could cross off.
How to borrow this for your own business
You do not need to be a bookkeeper for this pattern to work. Any solo business that depends on clients sending things (photos for a designer, measurements for a contractor, intake forms for a coach) can use the same three moves:
- Start from what each client actually sent last time, and let AI turn it into a personal checklist.
- Send the request early, as a heads-up rather than a demand.
- Remind with specifics: name what is missing, not what you want in general.
And keep the guardrails: no sensitive identifiers in the chat, a human read before anything goes out, and a plain sentence telling clients how you use AI.
What is the one thing you chase clients for every single year? Tell us in the comments, and we will suggest a checklist prompt for it.



