Three Years of ChatGPT: How AI Transformed Professional Services

Three years ago this month, OpenAI released ChatGPT to the world. It broke adoption records—hit 100 million users faster than any other app in history. And for the professional services world, it was a kind of alarm clock. Suddenly, generative AI wasn't theoretical anymore. It was in people's hands. It could write. It could analyze. It could reason through problems.

Let me walk through what actually changed in professional services in three years, what didn't, and what the lessons are as we head into 2026.

Year One (Late 2022–2023): The Shock and Hype

When ChatGPT launched, the reaction was binary. Some people thought it was the end of knowledge work. Some people thought it was an expensive toy. Both were wrong, but the wrong-ness was instructive.

In professional services, Year One was mostly experimentation. Partners were playing with ChatGPT on their laptops, writing some emails faster, drafting some initial proposals. But almost no firm deployed it seriously. Why? Because using ChatGPT for client work came with real risks: hallucinations, IP concerns, no audit trail, no governance. And enterprise adoption required getting past procurement, IT, legal, and compliance.

The real value that emerged in Year One was simpler: ChatGPT was a great brainstorming partner. A consultant stuck on a problem could ask ChatGPT for frameworks and get them in seconds. That was genuinely useful, even if it wasn't transformational.

Year Two (2023–2024): The Build Phase

Once the hype cooled, actual work started. Firms began building specialized AI applications. Legal firms built document analysis tools. Consulting firms built industry research assistants. Accounting firms built audit support systems. Not ChatGPT directly, but tools built on top of the same foundation models.

This is where real value started accruing. A law firm that trained an AI model on 10 years of contracts could now ask it questions about client agreements at scale. A consulting firm that built a client data assistant could let junior staff access deep knowledge without bothering a partner. An accounting firm that deployed an AI research tool could reduce time spent on standard research by 50%.

But Year Two also revealed the real challenges. Data quality mattered more than model capability. Governance was harder than expected. Change management was slower than hoped. Many projects that started with enthusiasm hit walls when it came time to actually integrate with legacy systems and organizational processes.

Year Three (2024–2025): The Maturation

By now, the firms that survived Year Two had working AI systems. They'd learned hard lessons about what worked and what didn't. And they'd realized that the AI tool itself was actually the smaller part of the problem. The bigger part was: How do you change how people work? How do you govern AI decisions? How do you stay compliant?

What we're seeing in Year Three is more realistic expectations. Firms aren't expecting AI to replace people. They're expecting it to make people more productive. They're not deploying bleeding-edge models. They're deploying stable, tested, somewhat boring models that work reliably.

The competitive advantage has shifted. It's no longer "Can you use AI?" Almost everyone can. It's "Can you use AI better than your competitors?" And that's about process, training, governance, and persistence—not about having a fancier model.

Three Things That Actually Mattered

1. The Capability Jump Was Real, But Limited

In 2022, everyone thought AI would automate away most knowledge work. Three years later, it's clear: AI is great at augmentation, much less great at full automation. A consultant with AI is 30–50% more productive. But the consultant is still required. This is less apocalyptic than the 2022 hype, but it's also more durable as an actual business model.

2. Data Quality Is More Important Than Model Capability

The firms that got the best results from AI weren't the ones that licensed the most expensive models. They were the ones that cleaned their data, organized their knowledge, and integrated their systems. That's unglamorous work. But it's foundational.

3. Governance Is Now Central to Competitive Advantage

The first movers on AI didn't know how to govern it properly. By Year Three, the firms that had figured out governance—who can use what model, what the AI can decide vs. what needs human review, how to audit AI decisions—those firms had a real advantage. Everyone else was playing catch-up on compliance and risk management.

What Didn't Change As Much As Expected

I expected pricing pressure. If AI can do X% of the work, shouldn't firms charge less? In practice, most firms haven't cut prices. Instead, they've raised quality expectations or worked on more complex problems. The productivity gain went to better solutions and higher margins, not lower prices. That's been surprising to some, but makes sense: client needs are elastic. If you can give them better quality or faster delivery, they'll pay.

I expected faster adoption. It's still slower than I thought three years ago. But slower adoption is fine—it means more sustainable implementation, less chaos, and less buyer's remorse.

What Comes Next

If Year One was shock, Year Two was building, and Year Three was maturation, what's Year Four?

I think we're entering an era of AI-augmented specialization. Firms won't compete on "we have AI." They'll compete on "we have AI+expertise in X that makes us exceptional." An accounting firm with an AI assistant that understands tax law is more valuable than an accounting firm with a generic AI assistant.

We're also moving toward AI agents—systems that can handle entire workflows, not just assist with steps. This requires more sophistication in governance and training, but the payoff is higher.

And we're seeing consolidation. The firms that invested seriously in AI in Years One and Two now have structural advantages. The firms that waited are starting to feel the gap. That gap will only widen.

The Real Lesson

If I could send a message to my 2022 self, I'd say: "You were right that AI would matter. You were right about the productivity gains. But you underestimated how much the social and organizational factors would matter, and how much the competitive advantage would go to the firms that were patient and systematic rather than aggressive and flashy."

ChatGPT didn't transform professional services overnight. But over three years, the accumulation of better tools, better process understanding, and better governance is genuinely transforming the firms that committed to it seriously.

Three years in, I'm more convinced than I was in 2022 that AI is durable and central to the future of professional services. But I'm also much more convinced that the winners won't be the ones who moved the fastest. They'll be the ones who moved the smartest.