This week's news from legal AI and the IRS points to a practical shift for professional-services firms: the work is moving toward systems that can handle longer, more structured tasks and produce clearer, more defensible outputs.
Harvey Tenet shows where legal AI is heading
Harvey's new Tenet model is positioned around long-horizon legal work, with a focus on agents that can move through complex tasks from start to finish. That matters for law firms because it is a sign that the market is moving beyond single-turn chat and toward workflow design.
For firm leaders, the real takeaway is not the model name. It is the operating model behind it: firms will increasingly need to decide which parts of a matter can be standardized, which need human review, and where custom models or agents can improve consistency without losing control.
IRS updates point to more structured compliance workflows
The IRS recently added a digitally authenticated Tax Compliance Report to individual online accounts, giving taxpayers a clearer way to show filing and payment status without disclosing unnecessary personal return information. That kind of output is exactly what clients expect from modern professional-services operations: targeted, easy to use, and privacy-aware.
The IRS also approved proposed guidance on Trump Accounts and created an Office of Conservation Easements to centralize technical expertise and coordinate strategy. Both developments suggest a broader trend toward more specialized administration, clearer rules, and more centralized case handling. Firms that support clients in these areas will benefit from workflows that can route work to the right specialist, assemble the right facts, and keep a clean record of what was reviewed.
What this means for custom AI and automation in firms
The common thread across these stories is the need for systems that do more than draft text. Law and accounting firms should be thinking about agentic workflows that can gather inputs, check for missing information, move through review steps, and produce a final package that a professional can sign off on.
That approach is especially relevant where the stakes are high: audit fraud procedures, tax compliance support, complex client intake, and regulatory responses. In those settings, custom AI should be built around controls, escalation paths, and evaluation rather than around generic chatbot convenience.
Build for the workflow, not the demo
For many firms, the best first step is not a broad AI rollout. It is choosing one repeatable workflow and defining what success looks like: what the system should collect, what it should flag, and where a lawyer or accountant must intervene.
As the news this week shows, the firms that win are likely to be the ones that treat AI as operating infrastructure. That means selecting use cases with clear business value, strong supervision, and a measurable review process.
- Focus custom AI on repeatable, high-value workflows rather than general-purpose chat.
- Use controls and review steps to keep agentic workflows defensible in client and regulatory settings.
- Match automation to the structure of the work: intake, compliance checks, document assembly, and escalation.
- Evaluate AI by output quality, completeness, and supervision fit, not just speed.
Sources watched
- Harvey Tenet, Nashville, Legal Innovators + (Artificial Lawyer)
- AICPA Auditing Board OKs New Standard Clarifying Auditors' Role in Detecting Fraud (CPA Practice Advisor AI)
- IRS Issues Proposed Regs on Eligible Investments for Trump Accounts (CPA Practice Advisor AI)
- IRS Adds Tax Compliance Report to Individual Online Accounts (CPA Practice Advisor AI)
- IRS to Open Office of Conservation Easements (CPA Practice Advisor AI)
