It's been six months since ChatGPT launched. Six months since the first "this will change everything" reactions. Now we have data. We have winners and losers. We have lessons.

Here's what I'm seeing in professional services.

The Winners

Firms that treated it as an operations tool, not a strategy tool. They asked "where does this actually make our team faster?" and ran pilots on real work. Now they're seeing 10-20% productivity gains on specific tasks.

Firms that moved fast and gave ownership. Appointed someone (usually operations) to lead it. Gave them budget and authority. They ran pilots, measured, and scaled. They're six months ahead.

Firms with good change management. They didn't just roll out ChatGPT. They explained why, trained their team, and set expectations. Adoption is natural, not forced.

Firms that stayed practical. No grand visions. No AI-first strategy. Just "we're using this for intake. For email drafting. For meeting summaries." Boring, but it works.

The Losers

Firms that wanted perfect compliance before starting. "We need a DPA, we need a policy, we need legal to sign off." Six months later, they still haven't run a pilot. Their competitors are ahead.

Firms that expected magic. "If we implement ChatGPT, we'll be 50% more productive." No specific plan. No measurement. No champion. Six months later, a tool nobody uses.

Firms that put IT in charge without operations. IT correctly said "we need data governance, security protocols, policy." But without someone focused on operations and business value, it became an IT initiative instead of a business initiative. Adoption suffered.

Firms waiting to see "what wins." Watching the market. Waiting for regulation to settle. Waiting for a better model. Meanwhile, the decision already happened: ChatGPT and similar tools are in. The question is how your firm uses them.

What I Learned

Speed matters more than perfection. The firm that started with a flawed pilot and refined it beats the firm that waited two months for perfect policy.

Ownership is everything. When one person (ops director, partner, manager) says "I'm responsible for this," it happens. When it's "everyone should figure this out," nothing happens.

Operations-first wins. Firms that started with operations improvements (client intake, email drafting, meeting notes) got wins fast. That created momentum. Firms that started with analytics or strategy got bogged down.

Team capability matters. Your team's ability to write good prompts, understand limitations, and apply judgment is as important as the tool itself.

Policy needs to enable, not restrict. The best policies don't say "don't use AI." They say "use AI for X, not for Y. Here's how to do it safely."

What Comes Next

Better models. GPT-4, Claude, Bard are all improving. GPT-5 is coming. The tools will get better and cheaper.

More integration. AI won't be a separate tool. It'll be built into your tools. Word, Excel, Slack, email. It'll just be how work gets done.

Specialization. Generic ChatGPT is useful, but specialized models for law, accounting, healthcare will be better.

Regulation. Rules will come. They'll be fragmented across jurisdictions. Firms that have thought about compliance early will adapt easily.

For Firms Starting Now (Late Adopters)

If you're just starting, you're not behind. You have the advantage of seeing what worked and what didn't.

Do: Start with operations. Pick a real bottleneck. Run a two-week pilot. Measure time saved.

Don't: Wait for perfect policy. Wait for the best tool. Wait for your competitors to force your hand.

Do: Appoint an owner. Give them budget. Give them authority.

Don't: Expect magic. Plan for 10%, not 50%.

The Retrospective

Six months in, ChatGPT isn't the revolution some predicted. It's also not the minor tool others dismissed. It's a significant productivity tool that works well when used thoughtfully.

The firms that understood that are winning. The firms that didn't are wondering why their AI initiative failed.

The lesson: AI is real, useful, and worth your time. But it requires the same discipline as any other business initiative. Clarity about what you're solving for. Ownership. Measurement. Iteration.

Get those right, and AI changes your firm. Get them wrong, and you have an expensive tool nobody uses.