Where Does Your Firm Fit on the AI Adoption Curve? The Thomson Reuters 2026 Benchmark.
Published October 2, 2025 · By The Crossing Report
Published: March 15, 2026 | By: The Crossing Report | 7 min read
Summary
Thomson Reuters CoCounsel has crossed 1 million users. GenAI adoption at the org level doubled in one year — from 22% to 40%. Only 15% of professional services firms currently use agentic AI, but 77% expect it to be central to their workflow by 2030. The question is no longer "is AI coming?" It's "where is your firm on the curve, and what does the next move look like?"
The Benchmark You've Been Waiting For
The Thomson Reuters 2026 AI in Professional Services Report draws on research across law, tax, accounting, audit, risk, and compliance professionals in dozens of countries. It's the most comprehensive benchmark available for professional services firm AI adoption, and the numbers are clearer than most coverage suggests.
The headline stat: GenAI org-level adoption doubled to 40% in one year. In 2025, roughly 1 in 5 professional services firms had formalized AI at the organization level. In 2026, it's 2 in 5.
The optimism gap: 66% of professionals feel optimistic about AI's future impact on their work. But 24% now see AI as a job threat — up from 15% last year. The worry is growing alongside the adoption. That tension is what this report captures: firms are moving forward, and some people are uncertain about what moving forward means for them.
The milestone: CoCounsel reached 1 million users across 107 countries, three years after launch. A quarter of Fortune 1000 companies now use it. The platform covers law, tax, accounting, audit, risk, and compliance — not just legal.
The Three Adoption Tiers
The TR data, combined with what we're seeing across the professional services market, maps cleanly to three tiers. Most firms sit clearly in one of them.
Tier 1: Experimenter
Who's here: The majority of small professional services firms. You've used AI tools — probably ChatGPT, possibly Copilot, maybe a purpose-built tool like Clio or August. But AI isn't embedded in a consistent workflow. Different people on your team use different tools, or no one uses them consistently. There's no firm-wide policy, no approved tools list, no training. You're curious but not systematic.
The data signal: The gap between "experimenting" and "formalizing" is large. TR found that 40% of firms have org-level GenAI adoption — meaning 60% have not crossed that line. Of the firms that have adopted, only 18% measure ROI. Most adoption is informal.
What's at stake: Shadow AI. When staff use unapproved AI tools with client data — which they are, whether you know it or not — your firm carries the professional liability for whatever those tools do. The experimenter tier is not a stable position in 2026.
Tier 2: Formalizer
Who's here: The 40% with org-level adoption. You have a written AI policy (or you're building one). You have a list of approved tools. You've done at least some staff training. AI is embedded in at least one workflow your team runs consistently — client meeting notes, document drafting, research, reporting.
The data signal: Moving from experimenter to formalizer is the biggest leverage point available to a small professional services firm in 2026. Firms with a written AI strategy outperform firms without one by 3x on ROI (Thomson Reuters 2026). That's not because the strategy is magic — it's because firms that document their approach are the ones that actually implement consistently.
What's at stake: The difference between AI that helps your firm and AI that creates liability exposure. At the Formalizer tier, you have a governance foundation. At the Experimenter tier, you're flying without instruments.
Tier 3: Agentic
Who's here: 15% of professional services firms currently. Agentic AI goes beyond assistance — it executes multi-step tasks autonomously. An AI agent can intake a matter, research applicable rules, draft a memo, flag review items, and update a client on status. The human reviews and approves; the AI handles the execution chain.
The data signal: 77% of professionals expect agentic AI to be central to their workflow by 2030. That's not a fringe view — it's the majority expectation across the industry. The gap between where the industry is today (15% agentic) and where it expects to be in four years (majority agentic) is the opportunity window.
What's at stake: This is where the competitive moat builds. Firms that deploy agentic workflows in specific high-volume service areas — tax prep, contract review, client onboarding, matter status reporting — will process more work per staff member. The ROI math changes structurally, not incrementally.
A Self-Diagnostic: Which Tier Are You In?
Answer these five questions honestly.
1. Does your firm have a written AI policy? If no → Experimenter tier. If yes → potential Formalizer or above.
2. Do you have a list of approved AI tools your team is allowed to use with client data? If no → Experimenter tier. The tools your team is using with client data right now may not be on your radar.
3. Is there at least one workflow in your firm where AI is used consistently by multiple team members — not just occasionally by one person? If no → still Experimenter. If yes → Formalizer threshold.
4. Have you ever measured whether an AI workflow saved time, reduced errors, or increased throughput? If no → you're with the 82% who don't measure. Whether Experimenter or Formalizer, add a measurement habit before your next AI decision.
5. Does any AI tool in your firm complete tasks autonomously — without a human doing each individual step? If yes → you're touching the Agentic tier. If no → that's fine for 2026. That's where most small firms should be.
The Optimist vs. Worried Split
TR found 66% of professionals are optimistic about AI's future impact on their work. 24% see it as a threat.
The characteristic that distinguishes the two groups is almost always whether they've had a successful AI experience — not a demo, a podcast, or a press release, but a workflow where AI actually made their specific work better. One successful experience tends to flip the emotional posture from threat to opportunity.
For a firm owner who still mostly feels worried rather than hopeful: the goal is to get one person in your firm one successful AI experience before the end of March. Not a training, not an evaluation — an actual piece of client work done faster or better with AI. That experience is the entry point. Everything else follows from it.
What "77% Expect Agentic AI to Be Central by 2030" Actually Means
It means the current 15% who use agentic AI today are not early adopters. They're on-ramp users. The main wave is expected within four years.
For a small firm owner, this is the timeline to plan around:
- 2026: Formalizer tier. Build the policy, the tools list, the first consistent workflow. Get your governance foundation in place.
- 2027–2028: Extend consistent workflows to two or three service areas. Measure ROI. Start identifying the first agentic candidate — the workflow where the steps are repetitive enough to hand to an agent.
- 2029–2030: Agentic deployment in specific service areas. The firms that built governance foundations in 2026 will be the ones who can safely move fast in 2029. The ones that skipped governance will be the ones managing a liability event instead.
The urgency is not "deploy an AI agent this week." The urgency is "don't still be an Experimenter-tier firm in 2027 when the Formalizer gap becomes a visible competitive disadvantage."
The One-Measurement Habit
Thomson Reuters found that only 18% of firms measure AI ROI. This is the single most valuable habit to add this month, before any new tool purchase or policy document.
Pick one workflow where AI is already in use. Measure three things:
- Time per task before AI and after
- Error or revision rate before AI and after
- Staff satisfaction with the workflow (1-5 scale, informal is fine)
Run the measurement for 30 days. Then make your next AI decision based on what you found — not based on what a vendor told you.
Firms that close the measurement loop are the ones that make better second decisions. The 18% who do this are capturing a compounding advantage.
The Action Item This Week
If you're at Experimenter tier: Do one thing. Write down the AI tools your team is currently using, including ones they use personally or informally. That's your starting list. The policy starts there.
If you're at Formalizer tier: Identify the first workflow candidate for a second AI embed. You have one consistent AI workflow. What's the second one? Pick the most repetitive, high-volume task that happens at least weekly. That's your next target.
If you're at Agentic tier: You're in the 15%. The question now is which service areas to extend to next and whether your governance framework is scaling with your agent deployment.
The Crossing Report covers the AI transition for professional services firm owners — accounting, law, consulting, staffing, and marketing agencies. Subscribe here for weekly field reports on what's changing and exactly what to do next.
Related reading:
Frequently Asked Questions
What does the Thomson Reuters 2026 AI in Professional Services Report say about adoption rates?
The Thomson Reuters 2026 AI in Professional Services Report, based on research across law, tax, accounting, and compliance professionals, found that GenAI org-level adoption doubled to 40% (from 22% in 2025). Only 15% currently use agentic AI — AI that acts autonomously rather than just assisting. But 77% of professionals expect agentic AI to be central to their workflow by 2030. Separately, CoCounsel crossed 1 million users in early 2026, signaling that large-scale formal adoption is underway at professional services firms.
What are the three AI adoption tiers for professional services firms?
Based on the Thomson Reuters 2026 data and observed firm behavior: (1) Experimenter — using AI tools individually or in isolated workflows, no firm-wide policy, no ROI measurement (the majority of firms today); (2) Formalizer — AI adoption formalized with a written policy, approved tools list, and training; this is the 40% with org-level GenAI adoption; (3) Agentic — AI handling multi-step autonomous workflows, AI agents completing tasks end-to-end with minimal human intervention; this is the 15% currently. Thomson Reuters found only 18% of firms measure AI ROI, meaning even many 'formalizers' are adopting without a feedback loop.
What is agentic AI and why does it matter for small professional services firms?
Agentic AI refers to AI systems that don't just generate outputs — they plan and execute multi-step tasks autonomously. Instead of helping you draft a document, an agentic AI tool could intake a client matter, research applicable statutes, draft a response, flag review items, and send a client update — with you reviewing the output rather than doing each step. Thomson Reuters found 77% of professionals expect this to be central to their workflow by 2030. For small firms, the implication is that the gap between Formalizer-tier and Agentic-tier firms will become a competitive moat within 3-4 years. The time to build workflow foundations is now, not 2029.
Should a small firm try to reach the 'agentic' tier now?
Not immediately. The Formalizer tier — having a written AI policy, an approved tools list, and at least one formalized workflow — is the urgent and achievable milestone for most 5-20 person firms in 2026. Once formalizer infrastructure is in place (documented policies, consistent tools, staff training), moving toward agentic workflows becomes a natural extension. Firms that skip the Formalizer tier and try to deploy autonomous AI agents without governance foundations create professional liability exposure without the oversight scaffolding to catch errors. Build the foundation first.
What does 'only 18% of firms measure AI ROI' mean in practice?
Thomson Reuters found that only 18% of professional services firms track whether their AI investment is actually paying off. Most firms adopt tools based on vendor claims, peer pressure, or fear of being left behind — and then never close the feedback loop. The practical consequence: firms don't know which AI workflows are saving time, which are adding risk, and which are sitting unused. The firms in the 18% who do measure ROI are making better second and third decisions about AI. The others are essentially flying blind. For a small firm owner, this is an easy gap to close: pick one workflow, establish a baseline (time spent, error rate, or client throughput), run AI-assisted for 30 days, measure the difference.