The Inbox and the Document Chase: AI for Accountants and Professional Services
In a small practice, the skilled work is rarely the hard part. Getting the right documents from the right client before the deadline is. Here's how AI sorts the inbox, chases what's missing and keeps deadlines in view.
Daniel Thomson4 Nov 2026 · 5 min read
If you work in a small accounting practice, law firm, mortgage brokerage or consultancy, you know the shape of the week. The skilled work, the part clients actually pay for, gets squeezed between two other jobs: reading email and chasing paperwork.
One client sends three emails about one question. Another sends bank statements, but not the right months. A third replies to a thread from last year with a new question at the bottom. Someone has to read it all, work out what it means, file it in the right place and reply.
Where the time goes
The cost isn't in any single email. It's in the switching.
- Triage. Every message has to be read just to work out who it's for and whether it's urgent.
- The document chase. A job can't start until the paperwork is in, and the paperwork arrives in pieces: by email, by phone photo, or by "I'll drop it in next week".
- Long threads. When a client rings, someone has to scroll back through months of email to find out where things got to.
- Deadlines. Due dates live in a spreadsheet, a practice tool and someone's head, and they don't always agree.
What the AI does, step by step
Here's the same inbox with AI teammates doing the first pass: an office manager for email and a document clerk for the paperwork.
- Every email is read and sorted. Each message is matched to the right client and job, tagged by type (question, documents, request, complaint) and given a priority with a short reason. Attachments are filed against the job.
- Documents are checked against a list. For each type of job, you keep a checklist of what you need. As documents arrive, the AI ticks off what's in and notes what's still missing, including the wrong-months problem.
- The chase is drafted. For anything outstanding, the AI drafts a polite, specific request: "Thanks for the March to June statements. We still need July and August for the business account." Specific requests get answered faster than vague ones.
- Threads are summarised. Open any client and you see a short summary of where things are, what was agreed and what's waiting on whom, with links back to the original emails.
- Replies are drafted. Routine questions get a draft answer in your firm's tone, based on the client's file. Anything that needs judgement is flagged for a person, not guessed at.
- Deadlines stay in view. Due dates are pulled into one calendar, with reminders to the team well before they bite and a list of jobs at risk because documents are still missing.
Where a person stays in control
Professional advice carries your name and your liability, so the line is clear.
The AI drafts. A person decides. No reply, chase or piece of advice goes to a client until someone on the team approves it. The AI doesn't sign off work, lodge returns or make payments.
Every action is logged: what was filed, what was drafted and who approved it. If an email lands against the wrong client, you can see it and undo it. That record is also useful when a client asks, months later, "did I send you that?"
What changes day to day
Think of a typical five-person accounting practice with a few hundred clients and a busy season that takes over every inbox. It's an illustration, not a real firm, but the shape will be familiar.
Before, the first hour of every day went on reading and sorting. Chasing documents was a job nobody owned, so it happened in bursts, usually late. One partner kept the deadlines in their head.
With AI doing the first pass, each person starts the day with a sorted list of what's theirs. The chase emails are drafted and waiting, and approving a batch takes minutes. The risk list shows which jobs will miss a deadline if documents don't arrive this week, while there's still time to do something about it.
When a client rings, whoever answers can read the thread summary in thirty seconds instead of scrolling for five minutes.
How to start small
The inbox is the right place to start, because it touches everything else.
- Connect your Gmail or Microsoft 365 inbox.
- Let the AI sort and summarise for a couple of weeks without drafting anything.
- Once the sorting is right, turn on drafted replies and document chases, with every one approved by a person.
Add deadline tracking next, then document checklists for your most common job types.
If you'd rather learn to use these tools yourself first, we also run one-to-one AI training, built around your own inbox and documents.
The short version
AI won't give your clients advice, and it shouldn't. What it can do is take the reading, sorting and chasing off your team, so skilled people spend their day on skilled work.
If that sounds like your inbox, describe your business on our home page and see a first plan for your own system. Or get in touch and we'll talk it through.
Daniel Thomson runs Ai Tearoa, a Christchurch company that builds small businesses their own system, with AI teammates that do the admin and a person who approves anything that goes out. The practice in this post is an illustration, not a customer story.
