Recent AI news points in the same direction: work tools are becoming more agent-like, and the value is shifting from generic chat to systems that can help with defined office tasks. For law and accounting firm leaders, that makes this a good moment to rethink where AI can safely support intake, follow-up, internal coordination, and other repeatable workflows.
AI is moving from chat to work execution
The latest AI headlines show a clear shift toward tools that do office work, not just answer questions. One report describes ChatGPT becoming a work agent, while another says Cursor is building an agent aimed at everyday office tasks like email and spreadsheets.
For professional-services firms, that matters because the useful unit of AI is no longer the chatbot. It is the workflow: a defined sequence of steps where AI can help triage, draft, route, summarize, or prepare information for a human to review.
Why custom AI beats a generic assistant
Generic AI can be helpful, but firms usually get better results when the system is built around a specific job. That is especially true in law and accounting, where accuracy, supervision, and client context matter.
A custom workflow can be designed around one process at a time, such as client intake, matter or engagement triage, document summarization, or follow-up coordination. This keeps the AI narrow enough to manage and easier for teams to trust.
Supervision and control still matter
Another news item highlights concerns about benchmark quality and the need to move beyond surface-level evaluations. For firms, the lesson is practical: do not adopt AI based on demos alone.
Any agentic workflow should have clear guardrails, human review points, and simple rules for when the system should stop and ask for help. That is especially important where client data, legal judgment, tax positions, or deadlines are involved.
Where law and accounting firms should start
The best first projects are usually the ones with repetitive inputs and predictable outputs. That may include routing new leads, collecting intake details, organizing documents, creating first-pass summaries, or preparing internal task lists.
Firms should also look at the handoffs between people and systems. Agentic AI is most useful when it reduces friction between email, forms, documents, CRM, and practice or workflow tools.
Succession, training, and process design are part of the same problem
A separate CPA practice article on succession planning makes a related point: firms should not rely on assumptions about who wants more responsibility. That is a reminder that technology should support real operational planning, not replace it.
The same discipline applies to AI. Before building an agent, firms should define the process, the owner, the review step, and the outcome they want. Then they can map the technology to the business need instead of the other way around.
- Start with one narrow workflow, not a broad AI rollout.
- Build human review into every agentic process that touches client work.
- Use custom AI to improve handoffs, intake, and internal coordination.
- Treat workflow design and succession planning as part of the same operating discipline.
Sources watched
- Price Wars are Nigh, SensorFM, Harness Matter, & More (The White Box)
- State CPA Society News & Updates - July 2026 (CPA Practice Advisor AI)
- ChatGPT Just Became A Work Agent (AI Daily Brief)
- Michigan Tax Preparation Business Owner Charged with Fraud (CPA Practice Advisor AI)
- Avoid Assumptions: Use Career Conversations to Guide Your Firm's Succession Planning (CPA Practice Advisor AI)
