The number is real. AI adoption across professional services organizations in the US almost doubled in a single year, from 22% in 2025 to 40% in 2026. That's not a slow creep. That's a real inflection point. And yet the majority of those same organizations can't tell you whether it's actually working.
TL;DR
- AI adoption across US organizations nearly doubled from 22% to 40% between 2025 and 2026, per the Thomson Reuters AI and Professional Services Report 2026.
- Only 18% of those organizations are actually tracking ROI on the AI tools they're using.
- Most "ROI metrics" are employee usage and satisfaction... not revenue, new clients, or outcomes with a clear line to the bottom line.
- Legal and accounting firms are the most hesitant to adopt, with 35% of law firms reporting no plans to use AI at all.
- The real edge isn't just using AI. It's getting your whole team on a shared system where context compounds over time.
The Number That Jumped Out
Two thirds of respondents to the Thomson Reuters survey said they feel excited or hopeful about generative AI in their industry. That's a meaningful shift from even a year ago when the conversation was still "will this stick?"
It's sticking.
AI is going to be part of the work the same way Excel is part of the work. The same way Slack or Teams is. You log in, you use it, it's just the interface now. What's interesting isn't whether organizations are adopting it. It's whether they know what they're getting from it.
And the answer, mostly, is no.
Only 18% Are Tracking ROI... and Most Are Measuring the Wrong Things
Only 18% of organizations say they're actively measuring the ROI of their AI tools. That's a low number. But what's more telling is how the ones who are measuring are doing it.
The top metrics being used:
- Internal cost savings (77%) — this one makes sense
- Employee usage — just because people are using it doesn't mean you're getting value
- Employee satisfaction — okay
- New business won — only 17% are tracking this
- Client satisfaction — only 26%
Think about that for a second. The most common ROI metric for AI is... whether employees use it. Not whether clients are happier. Not whether you won more deals. Whether people logged in.
That's not ROI. That's adoption tracking dressed up as impact.
The organizations doing it best are corporate tax and corporate legal departments, where 25–26% report actually measuring their AI ROI. Everyone else is mostly guessing or not measuring at all. Government? 73% said they don't even know whether they're measuring it. (That checks out.)
If you want to actually know where to start when it comes to AI strategy, the small business strategy diagnosis quiz can help you figure out where the real bottleneck is before you start stacking tools on top of a process that hasn't been proven yet.
Why Communication Breakdown Is the Real Problem
Here's what I think is happening: management is trusting that employees will use AI and that productivity will follow. But they're not building the infrastructure to guide it.
If your business has an IT department, you need someone sitting with employees, watching how they actually use AI, and then approving the right tools for the right jobs. Not just "use ChatGPT if you want." That approach gets you a bunch of one-off prompts, no shared context, and no way to measure anything.
The businesses that are going to win are the ones that get their whole team on one system, saving context and sharing knowledge over time.
Here's the version I want you to imagine: you log into your AI and it knows what everyone on your team did yesterday. You can pick up your own work right where you left it... or seamlessly step into someone else's. That's not science fiction. That's what shared organizational AI context looks like when it's done right.
Legal and Accounting Firms: Hesitant for Real Reasons
The report gets interesting when it digs into professional services specifically. 35% of law firms say they have no plans to use AI at all. And there's actually some logic to the reluctance.
The confusion is real. 40% of firm respondents said they had received instructions from clients both to use AI on matters and not to use AI on matters. Mixed signals everywhere. And about three quarters of both corporate respondents and firm respondents agreed that firms should be taking the lead on starting these conversations... but firms are hesitant, citing concerns about quality and fidelity.
The honest subtext is that some of these firms feel threatened. And they should, a little. The percentage of lawyers calling AI a major threat to unauthorized practice of law rose from 30% in 2025 to 50% in 2026. That's not paranoia, that's a real legal and ethical question that the profession hasn't worked out yet.
The hourly billing model is a compounding problem here. If you charge $500/hour and AI just compressed three days of paralegal research into three minutes... your revenue has a problem. The incentive structure actively works against adoption.
Jackson put it well on the show: right now very few clients trust that AI produces quality work. So the firm not using AI looks safer. But eventually that flips. And when it does, the firm that isn't using AI will be the one that takes longer to solve your problem.
What to Actually Do With This Data
The report is a useful snapshot. Most professionals now believe AI will make a major impact on jobs, billing, and revenue in their industries. That's a pretty wide acknowledgment of what's coming.
But acknowledgment isn't a strategy.
The businesses that are going to pull ahead aren't the ones that adopt the most tools. They're the ones that:
- Get their team on a shared system (not 12 different individual subscriptions)
- Pick one or two high-ROI use cases and actually measure them (revenue or cost savings, not vibes)
- Have someone internal staying two steps ahead of the employees so tools are approved and consistent
The AI tools checklist at infacto.digital/presentations/ai-tools-checklist.html is a good starting point for figuring out which tools are worth your attention and which ones are just noise.
The Last Adoption Wave You'll See Before It's Just Expected
Jackson said something on this episode that stuck with me. He said this data will look like a snapshot of history in ten years. The same way you could have asked in 1995 whether businesses planned to use the internet.
He's right.
The organizations at 40% adoption and no ROI tracking aren't behind because they're slow. They're behind because they skipped the infrastructure step. They added tools without adding standards.
The ones who build the infrastructure now... shared systems, clear metrics, approved tools, someone staying ahead of the users... those are the ones who will look back and say they got ahead of this.