UpTrajectory Review

Thomson Reuters has pushed a major update to CoCounsel Tax, Audit & Accounting, positioning the platform as a fully agentic AI system for practitioners. The original version launched last year, but this release signals a shift from AI as a drafting assistant to AI as an autonomous agent capable of executing multi-step workflows with minimal human prompting. For context, Thomson Reuters has been under pressure from both Big Four firms building proprietary AI stacks and a wave of nimble startups targeting the accounting and tax workflow space. Agentic AI — systems that can plan, reason, and act across tools rather than simply respond to queries — has been the dominant theme in enterprise software over the past year, and Thomson Reuters is clearly betting that tax and audit work is ready for that leap.

For small and mid-sized firms, this matters more than the press release framing suggests. Large firms have the capital to experiment with AI pilots and absorb failures; smaller practices do not. If CoCounsel's agentic capabilities genuinely work — meaning it can pull client data, reconcile figures, flag anomalies, and draft deliverables with reliable accuracy — it could compress the staffing gap that has plagued firms since the pandemic-era talent shortage. A three-person shop could theoretically handle the volume of a six-person practice. But that is a big if, and the cost of a wrong number or a missed filing in tax work is not a shrug-and-iterate moment. It is a liability.

What is genuinely new here is the agentic framing itself. Last year's release was positioned as an AI assistant; this one is being marketed as a platform that acts. That is a meaningful distinction, but it is also where we are skeptical. Thomson Reuters has a long history of rebranding incremental improvements as transformational, and the accounting profession has a long history of adopting new tools slowly and grudgingly. The under-reported question is not whether the technology works in a demo environment — it is whether it holds up when a client hands over a shoebox of receipts in April. We would want to see independent benchmarks, not vendor case studies, before advising any firm to restructure workflows around this.

The second-order effects are worth watching carefully. If agentic AI genuinely reduces the hours required for routine tax prep and audit support, the traditional billable-hour model faces real pressure. Firms that price on value rather than time will adapt; firms still billing by the hour may find their margins squeezed as clients question why a return that took six hours last year now takes two. There is also a competitive dynamic at play: firms that adopt early and market their AI-enhanced capabilities could pull clients from slower-moving competitors, particularly among younger business owners who expect tech-forward service. Conversely, firms that over-automate too fast and produce errors will face reputational damage that no efficiency gain can offset.

Our advice to practitioners: do not rip and replace anything this tax season. Instead, run CoCounsel or a competitor in parallel on a subset of low-risk engagements and compare outputs against your team's work. Watch for pricing changes — Thomson Reuters has a history of bundling AI features into premium tiers that quietly inflate subscription costs. Also watch how state boards of accountancy and the IRS begin to respond to agentic AI in practice; regulatory guidance on AI-assisted tax preparation is still nascent, and the liability question of who is responsible when an AI agent makes an error has not been settled. The firms that win here will be the ones that treat AI as a junior associate who needs supervision, not a partner who can be left alone with a client.

Takeaway: Pilot agentic AI on low-risk engagements before restructuring workflows, and treat it like a junior associate who needs supervision, not a replacement for professional judgment.