What Your Team’s AI Experiments Are Actually Teaching You

A few months ago, I started asking a simple question at every conference and event we run at Converge360: “How many of you have an official AI pilot underway?” Hands go up. Then I ask: “How many of you have employees already using AI tools on their own, outside of any official program?” Almost the same hands, and sometimes more.

That gap is the most important data point in enterprise AI right now.

Your team is already experimenting. The question isn’t whether the learning is happening, it’s whether you’re capturing it.

Here’s what I’ve observed from talking to IT leaders, practitioners, and business users across our communities: the most valuable AI education happening in organizations today isn’t in the training courses. It’s in the informal, often untracked experiments your people are running on their own. The marketer who figured out how to use Claude to cut her first-draft time in half. The developer who discovered that Copilot is great for boilerplate but needs careful review on complex logic. The analyst who learned (the hard way) that AI confidently hallucinates citations.

These are not failures; they are your organization’s emerging body of knowledge about AI, and most companies have no mechanism for capturing or spreading it.

That instinct has been validated over and over. The teams winning with AI right now aren't the ones that spent six months on a strategy. They're the ones who got their people using tools, watched what happened and kept adjusting.

When I look at what organizations genuinely learn from early AI experiments, it falls into four categories:

1. Where the real leverage is. Every team discovers this slightly differently. For some, it’s summarization and synthesis, for others, it’s that first-draft generation that’s needed, while some use it for code review or documentation. The experimentation really reveals where AI multiplies your people’s output, and that answer is almost always more specific and more interesting than the general promises in the vendor pitch.
2. Where the guardrails need to be. AI tools will occasionally produce plausible, wrong output, but they will do so confidently. Your team learns this viscerally through small experiments — which is exactly the right place to learn it, before mistakes that matter are made at scale. The team that discovers a hallucination problem in a low-stakes content draft is much better prepared than the team that discovers it in a client deliverable.
3.   What “good prompting” actually means in your context. Prompting is a skill, and it’s domain-specific. What works for a software engineer doesn’t work the same way for a finance analyst. Your early experimenters are building this muscle. The organizations that spread those learnings, through prompt libraries, internal wikis, and lunch-and-learns, accelerate the whole company.
4.   How resistant (or ready) your culture actually is. This is the one nobody talks about. AI experiments surface cultural attitudes toward change, automation, and risk. Some teams lean in immediately, others push back. Understanding human dynamics is almost as important as understanding the technology. You can’t see that from a readiness assessment; you see it from watching real people use real tools on real work.

The difference between organizations that compound their AI advantage and those who don’t comes down to one thing: whether they’ve built a loop between individual experiments and collective knowledge.

It doesn’t have to be complicated. A shared Slack channel where people post what’s working and what isn’t, a monthly 30-minute all-hands where someone demos something new they tried or a simple internal template: “I tried X, here’s what happened, here’s what I’d do differently.” These are not sophisticated knowledge management systems; they’re just making the invisible visible.

The companies I’m most bullish on right now aren’t the ones with the biggest AI budgets. They’re the ones that have created a culture where experimentation is expected, learning is shared, and iteration is continuous. That culture, once built, is genuinely hard to replicate.

If you’re a leader reading this, here’s the most practical thing I can suggest: find the people in your organization who are already using AI tools on their own and ask them what they’ve learned. Don’t make it a compliance conversation. Make it a learning-exchange and knowledge-share one.
You’ll be pleasantly surprised by what’s already in the room.

Posted by Daniel LaBianca on 05/21/2026