The latest news from AI and professional-services publications points in the same direction: firms do not get lasting value from AI just by turning it on. They get value when they redesign work so the system can operate with the right context, controls, and review paths.
AI gains are still getting lost in manual follow-up
AI Daily Brief highlighted a problem called bot sitting: the hidden work of feeding AI context, checking outputs, rerunning prompts, and cleaning up mistakes. The report cited in that summary says many digital workers are using AI and saving time, yet organizations are not seeing the same level of improvement.
For firm leaders, that is a useful warning. A custom AI tool that still needs constant supervision may help an individual employee, but it will not become a reliable firm workflow unless the surrounding process is built to reduce rework and ambiguity.
Custom AI needs more than a chatbot layer
The same pattern shows up in the accounting news. Intuit's latest platform updates were framed around accountant feedback and a view of accountants as human intelligence layered on top of the platform. That is a reminder that useful AI in firms is usually part of a broader operating model, not a standalone assistant.
For law and accounting firms, the practical takeaway is to start with a narrow workflow: intake, document triage, compliance review, matter setup, or tax support. Then define what the AI does, what the human reviews, and what data the system must pull before it can act with confidence.
Compliance and audit quality are becoming automation use cases
Smarsh's collaboration with AWS shows how compliance review workload can be reduced while preserving risk detection. That matters for regulated firms because it shows AI can support oversight functions when governance is designed into the workflow from the start.
The PCAOB's new firm consultation process also points to a broader theme: firms need clear interpretive paths when standards change. In practice, that makes a strong case for agentic workflows that route questions, capture review decisions, and keep a searchable record of how the firm applied its policies.
What this means for law and accounting firm operators
The common thread across these updates is that AI value comes from process design, not just software access. The firms most likely to benefit are the ones that treat AI as a workflow system with rules, checkpoints, and feedback loops.
That is especially true for custom AI. If your team is spending too much time correcting AI output, gathering missing context, or deciding when the tool can be trusted, the workflow is not yet ready. The opportunity is to redesign the work so the model does less guessing and the firm does less bot sitting.
- Choose one workflow where the firm already repeats the same review steps and automate that process end to end.
- Build AI with required context, escalation rules, and human review points so staff are not spending time cleaning up output.
- Treat compliance, audit quality, and documentation as part of the workflow design, not as after-the-fact controls.
Sources watched
- Bositting: The Work Draining AI Gains (AI Daily Brief)
- Intuit Connect ON: What You Need to Know About the Latest Platform Innovations (CPA Practice Advisor AI)
- Smarsh Collaborates with AWS to Reduce Compliance Review Workload by 77% (CPA Practice Advisor AI)
- PCAOB Unveils Firm Consultation Process to Drive Audit Quality Improvement (CPA Practice Advisor AI)
