6 min read
This is an illustrative story. Maya is a composite based on common situations solo bookkeepers describe, not a real individual, and the numbers are realistic estimates rather than reported results.
It is the second business day of the month, 9:40 PM, and Maya is on her third cup of tea. Fourteen small business clients, each with a slightly different mess. A landscaper who photographs receipts on the dashboard of his truck. A yoga studio owner who pays for everything with three different cards. A therapist who forwards invoices from her personal email “just in case.” Maya has run her solo bookkeeping practice for six years, and the first week of every month looks exactly like this.
By her own rough tracking, month-end close was eating around 40 hours spread across the first eight days of each month. That is a full working week, every month, spent mostly chasing paperwork rather than doing the part clients actually value: explaining what their numbers mean.
This is the story of how she cut that to about 20 hours without changing her accounting software, without hiring anyone, and without handing her clients’ financial judgment to a machine.
Where the 40 hours were really going
Before changing anything, Maya spent one month logging her time in fifteen minute blocks. The results surprised her. The actual bookkeeping, reconciling accounts and reviewing categories, took about 15 hours. The other 25 hours went to:
- Chasing missing documents from clients, usually by writing the same email over and over
- Retyping receipt details from blurry photos and forwarded emails
- Answering “quick questions” that arrived in four different apps
- Writing the monthly summary that each client received with their reports
“I thought I had a bookkeeping problem,” she would later tell a colleague. “I actually had a paperwork and writing problem.”
Step one: stop retyping receipts
The first change was the least glamorous and the most valuable. Maya moved every client onto a receipt capture app that uses AI to read a photo of a receipt and pull out the vendor, date, amount, and tax. Clients snap a picture or forward an email, and the details land in a shared queue for her to review.
She chose a tool similar to the one in our SparkReceipt review, mainly because clients found it easy. That mattered more than any feature list. A tool clients will not use saves nothing.
Crucially, she kept herself in the loop. Every extracted receipt still passes her eye before it touches the books. The AI does the typing; Maya does the checking. She also reminded clients that digital copies need to be kept properly, pointing them to the IRS guidance on recordkeeping so nobody assumed a photo in a camera roll was enough.
Time saved: roughly 8 hours a month.
Step two: let AI write the chasing emails
Maya’s second biggest time sink was asking for missing things. Each request needed to be polite, specific, and slightly different depending on the client’s personality and how many times she had already asked.
She built a simple routine using a general AI assistant such as ChatGPT or Claude. At the start of close, she pastes in a short list: client first name, what is missing, how many reminders already sent, and the tone that works for that person. The assistant drafts all fourteen emails in one go. She edits, then sends.
Two rules kept this safe. First, she never pastes full bank statements or account numbers into a general chatbot, only the minimum detail needed to write the email. Second, she reads every draft before it leaves. If you want a deeper version of this approach, our guide to chasing unpaid invoices with AI uses the same idea for collecting payments.
Time saved: roughly 6 hours a month.
Step three: one inbox for “quick questions”
Clients were texting, emailing, messaging on social media, and occasionally leaving voicemails. Maya set up a single intake form and a polite auto reply in every channel pointing clients to it. The form asks three questions: what is it about, how urgent is it, and is anything attached.
An AI step then sorts each submission into “needs Maya today,” “can wait until close,” or “already answered in your monthly notes,” and drafts a suggested reply. Maya still sends every answer, but she now handles questions in two focused blocks a day instead of dropping everything whenever her phone buzzes.
Time saved: roughly 4 hours a month, plus a level of calm that is harder to measure.
Step four: summaries clients actually read
Every client used to get the same thing: a profit and loss report and a balance sheet, with a two line note. Most never opened the attachments.
Now, after she finalizes the books, Maya gives an AI assistant the month’s key figures (no names or account numbers) and asks for a plain English summary in five bullet points: what changed, what looks unusual, and one thing to think about next month. She edits it heavily, because she knows each client’s situation, and the result is a short note clients genuinely read and reply to.
Some clients asked for more, so she started offering a simple visual dashboard as an upgrade, using an approach like the one in our guide to turning messy numbers into a live dashboard.
Time saved: roughly 2 hours a month, and two clients moved to a higher priced advisory package.
What did not work
Not every experiment paid off, and that is worth saying out loud.
- Letting AI categorize transactions unsupervised produced confident mistakes, such as treating a client’s personal grocery run as a business meal. Maya went back to reviewing every category, using AI suggestions only as a first guess.
- Asking a chatbot tax questions gave answers that sounded authoritative but were sometimes out of date. She now uses official sources and her professional judgment for anything tax related.
- Automating too many client emails at once made a few long time clients feel processed. She now writes the important conversations herself.
The result, and what it means for you
Adding it up, Maya went from around 40 hours to around 20 hours of month-end work. She did not use those hours to take on twice as many clients. She used about half for deeper advisory conversations with existing clients, and the rest she simply took back. As she puts it in the story, “I finally have a first week of the month that does not feel like a siege.”
You do not have to be a bookkeeper to borrow her approach. The pattern works for almost any solo service business:
- Log your time for one month before buying any tool. Find out where the hours really go.
- Automate typing and drafting first, because they are low risk and high volume.
- Keep judgment human, especially anything involving money, taxes, or a client relationship.
- Protect sensitive data by sharing the minimum with AI tools, a theme we cover in 5 Things Solo Owners Get Wrong About Putting Business Data Into AI.
For broader financial housekeeping beyond month-end, the US Small Business Administration’s guide to managing your finances is a solid, free reference.
If you logged your own time this month, which task would surprise you most?
