Recent news across accounting, tax administration, and legal AI points to a practical shift: firms are moving past experimentation and toward workflows that can handle real work with better oversight, fewer manual bottlenecks, and tighter controls.
Why better AI models matter for firm workflows
OpenAI's GPT-6 Astra is being positioned as a more capable model for end-to-end tasks, and coverage around it focused on whether models have crossed a dependability threshold that makes them useful for professional work. That matters for firms because the value of AI is not the chat experience itself, but whether it can support work from intake through review without creating new risk.
For law and accounting leaders, the main question is no longer whether AI can draft or summarize. It is whether a custom workflow can complete a defined task reliably, with the right handoffs, validation steps, and supervision built in.
Paper-heavy processes still create delays and cost firms time
The IRS paper return update is a reminder that manual and paper-based processes remain slow and fragile. The GAO review found paper returns took longer to process this filing season, and staffing and technology problems affected taxpayers who filed on paper.
That is the same operating problem many firms face internally: work gets stuck when it depends on repeated handoffs, missing information, or slow follow-up. Custom automation can help triage inbound documents, route exceptions, and keep routine matters moving while staff focus on higher-value review.
Fraud risk is also an automation design problem
The Connecticut nonprofit case shows how a shortage in financial records can sit inside routine processing until someone investigates. The allegations involved incoming funds, a discrepancy discovered during review, and additional charges tied to computer crime and identity theft.
For firm owners, the practical lesson is that automation should not only speed work up. It should also create better visibility: approvals, reconciliation flags, exception queues, and audit trails that make unusual activity easier to spot before it becomes a larger loss.
What professional-services firms should build next
The strongest use case here is not a generic chatbot. It is a custom AI workflow that takes in documents, classifies the matter, checks for missing fields, flags risk, and sends only the right items to staff for review.
For accounting firms, that could mean intake and document triage tied to exception handling. For law firms, it could mean controlled drafting and matter routing with vendor and privilege diligence built in. The common thread is the same: fewer manual bottlenecks, more standardization, and clearer oversight.
- Use AI for defined workflows, not open-ended chat, when the work has compliance or financial risk.
- Design automation to surface exceptions and audit trails, not just to save time.
- Start with one high-friction process such as intake, document triage, or reconciliation.
- Choose models and vendors based on reliability, control, and reviewability, not novelty.
Sources watched
- State CPA Society News & Updates - Sept. 2026 (CPA Practice Advisor AI)
- Harvey + Legora on OpenAI’s GPT-6 Astra (Artificial Lawyer)
- Former Connecticut Nonprofit Accountant Accused of Embezzling Over $54,000 (CPA Practice Advisor AI)
- IRS Processing of Paper Returns Took Twice as Long This Past Filing Season (CPA Practice Advisor AI)
