AI for Accountants Capstone Exercise — AI for Accountants

What you'll learn

Apply skills from across this course in a single realistic multi-part accounting scenario
Produce four concrete professional deliverables using AI tools with appropriate verification applied to each
Self-assess your output against the professional quality criteria covered in lessons 2 through 9

Across this course you have worked through the AI applicability tiers in accounting, AI-assisted bookkeeping and reconciliation, month-end close acceleration, audit workpaper preparation, tax research with appropriate verification, client communications, practice management workflow, data quality and verification, and the ethics and governance framework. Each of those skills addresses a distinct part of accounting practice. This capstone brings them together in a single scenario that mirrors the kind of situation a practising accountant actually encounters: a new engagement, disorganised starting materials, multiple tasks to complete, and a professional obligation to do all of it to a quality standard that you would be prepared to defend.

Capstone Exercise

New E-Commerce Client Engagement: Bookkeeping Triage, Close Checklist, Client Communication, and Tax Position Note

Context

You are a sole practitioner accountant managing 60 SME clients. A new client has just engaged you: a three-year-old UK e-commerce business selling homeware products, currently generating approximately £480,000 in annual revenue. The business has no prior accountant. The bank feed shows 18 months of transaction history in Xero, but the categorisation is inconsistent: personal expenses appear in business accounts, several recurring supplier payments are miscategorised, software subscription costs have been coded outside the scope of VAT, and there are no working papers from the prior period. The client has asked you to bring the books up to date, prepare the current year accounts, and advise whether the business might be eligible for an R&D tax credit related to custom product packaging software they commissioned from a developer.

Your Task

Complete four components using AI tools. First, categorise the bookkeeping risk: review the types of transactions described in the scenario and prepare a written bookkeeping risk assessment that identifies the three highest-risk transaction categories, explains why each is high-risk, and describes the review protocol you would apply to each using the approach from Lesson 2. Second, prepare a month-end close checklist: using an AI tool and the prompt approach from Lesson 3, generate a first-draft close checklist appropriate for a UK VAT-registered e-commerce business. Review and annotate the checklist, noting any items the AI included that are not applicable and any items specific to this client that the AI missed. Third, draft a client email: using the briefing approach from Lesson 6, draft an email to the client explaining what you have found in the initial bookkeeping review, what the client needs to provide to complete the accounts, and the professional tone appropriate for a new client relationship. Fourth, prepare a tax position note: using the research workflow from Lesson 5, research the basic eligibility criteria for the UK R&D tax credit scheme (RDEC or SME scheme as applicable for this client's size) as it applies to software development expenditure. Prepare a one-page tax position note that states what you found, identifies any areas where the AI research requires verification against primary sources, and notes any limitations on the position you can confirm without further information from the client.

Your notes (optional)

Deliverable

Four written deliverables: (1) a bookkeeping risk assessment identifying the three highest-risk transaction categories with review protocols for each; (2) an annotated close checklist for the client with notes on AI-generated items reviewed and any client-specific additions; (3) a client email covering the initial bookkeeping findings, information requests, and next steps; (4) a one-page R&D tax position note with the eligibility framework, key statutory references verified against primary sources, and explicit flags for any areas requiring further client information or primary source verification.

Quick check

The capstone asks for four AI-assisted deliverables from one engagement. What does it say about the verification each of them needs?

Select one answer.

Key takeaways

AI is most valuable in accounting practice when applied systematically across a full engagement: bookkeeping triage, close preparation, client communication, and research all benefit from structured AI assistance, but each requires a different verification standard matched to its professional consequences.
The bookkeeping risk assessment forces the skill from Lesson 2 to be applied proactively at engagement start rather than reactively at quarter-end: identifying high-risk transaction categories before the coding errors compound is the more professional and more efficient approach.
The close checklist exercise demonstrates the AI limitation described in Lesson 3: AI generates a good structural starting point from context, but client-specific risks that you know from the engagement (prior year gaps, personal transactions, inconsistent categorisation) require human annotation to appear in the checklist.
The client email exercise reinforces the Lesson 6 principle: the brief quality determines the output quality. A vague brief produces a generic email that could go to any client. A specific brief that describes what you actually found produces a letter that demonstrates professional competence and builds the new client relationship.
The R&D tax position note applies the full Lesson 5 workflow: AI orientation research, statutory reference verification against primary sources, explicit uncertainty flagging, and a clear statement of what you know versus what requires more information. This is the professional research standard, not a shortcut.