This is an illustrative scenario, not a client case study. It walks through how a WhatsApp chatbot for CA firms could take over the most repetitive job of the income tax return season: collecting documents from individual clients. There is no real firm behind it and no results are claimed. The aim is to show a business owner what such a system would do step by step, what it needs, what can go wrong and how you would know whether it was working.
The scenario
Picture a chartered accountancy practice with a few partners, a small team of article assistants and several hundred individual clients who file an income tax return each year. Every season the same thing happens. Staff message each client asking for Form 16, bank statements, interest certificates, capital gains statements and rent receipts. Clients send half of them, in several messages, over several weeks. Someone keeps a spreadsheet of who has sent what. The technical work of preparing returns is squeezed into the final days because documents arrive late.
The firm already talks to clients on WhatsApp, informally, from staff phones. The question is whether that conversation can be structured and partly automated.
How the WhatsApp chatbot for CA firms would work
- Set up the checklist. Each client is tagged with a profile, such as salaried, salaried with capital gains, or business income. Each profile has a document checklist that the firm defines.
- Get consent and open the conversation. The client opts in, for example by scanning a QR code on the engagement letter or tapping a link, and sends the first message. The bot replies with the list of documents needed for that client.
- Receive documents. The client sends photos or PDFs in any order. A document AI step identifies what each file is, for example a Form 16 or a bank statement, and ticks it off the checklist. If a file is unreadable or is the wrong year, the bot says so and asks again.
- Answer routine questions. Clients ask where to download a statement or what a particular form is. The bot answers only from content the firm has approved.
- Send reminders. Clients with pending items get a scheduled reminder listing exactly what is missing. Reminders stop when the checklist is complete.
- Hand over to a person. Anything involving judgement, such as a question about a deduction, a notice or fees, goes to the assigned staff member with a summary of the conversation so far.
- File and report. Documents are saved to the client's folder under a consistent naming scheme. Partners see a status view: complete, partly complete, not started.
Nothing in this flow prepares a return or gives tax advice. It only moves documents from the client to the firm and shows who is still pending.
Data and integrations it needs
- A WhatsApp Business Platform account on the official API, with a verified business and a dedicated number.
- A client list with name, mobile number, profile and assigned staff member. A spreadsheet export is enough to start.
- The firm's document checklists and its approved answers to common questions.
- Storage the firm controls, such as a document management system or a structured cloud drive, with access limited by role.
- A shared inbox where staff pick up handed-over conversations.
- A record of each client's consent and a simple way for them to opt out.
Integration with practice management software is useful but optional at the start. It depends on whether that software offers an API or only exports.
What WhatsApp itself charges
Messaging fees are separate from the cost of building the bot. According to Meta's pricing documentation, businesses have been charged per message since 1 July 2025. Template messages are priced by category and country. Non-template replies are free inside the customer service window, which opens when a client messages you and closes 24 hours after their last message, and utility templates delivered inside an open window are also free.
The design consequence is practical. If the client starts the conversation and uploads documents in one sitting, most of the exchange costs nothing in messaging fees. Reminders sent days later fall outside the window and are charged as templates, so it pays to send one clear reminder with the full pending list instead of several short ones. Check the current rate card before budgeting.
Realistic effort
As a planning estimate, not a quotation, a focused first version would take roughly four to seven weeks:
- Weeks one to three. Number setup and business verification, message templates submitted for Meta's review, checklist flows, document receipt and storage.
- Following two to four weeks. Document identification, the staff handover inbox, the status view and testing with a small group of friendly clients.
What extends this: integration with practice software, support for several languages, and stricter security requirements. What shortens it: starting with one client profile and one return type.
Risks and how to handle them
| Risk | How to reduce it |
|---|---|
| A document is identified wrongly | Use confidence scores, and have staff confirm each file when they begin the return |
| Sensitive financial documents are exposed | Role-based access, a retention policy, and a secure upload link for clients who prefer not to use WhatsApp |
| The bot drifts into tax advice | Restrict answers to approved content and route everything else to a person |
| Clients feel spammed | Message only clients who opted in, cap reminder frequency and honour opt-outs at once |
| Professional conduct concerns | Confirm that message wording and outreach comply with your professional body's rules before launch |
What to measure
- Share of clients with a complete document set by your internal cut-off date.
- Average days from first request to complete set.
- Staff hours spent chasing documents each week.
- Share of conversations that needed a human.
- Document identification errors found during return preparation.
- Opt-outs and complaints.
Record the first three for the previous season before you start. Without a baseline you cannot tell whether the system helped.
Next steps
If this scenario resembles your practice, our WhatsApp AI chatbot development page explains how we set up the official API, design conversations and connect them to your systems. For background, read how WhatsApp AI automation works and our guide on extracting data from documents automatically.
Frequently asked questions
Is this a real project you delivered?
No. It is a worked example written to show how the pieces would fit together. Any real project would start with your own client list, checklists and systems.
Can the chatbot give clients tax advice?
It should not. In this design it collects documents and answers routine process questions from firm-approved content. Questions that need professional judgement go to a member of staff.
Do clients have to install anything?
No. They use WhatsApp as they already do. The firm's side runs on the WhatsApp Business Platform.
Does WhatsApp automation for accountants work with our existing software?
It depends on the software. If it has an API, client and status data can sync automatically. If not, a spreadsheet import and export is a workable starting point.
