Every AI session I sit through feels like either a vendor pitch or a consulting firm's lead-gen — where do I go to hear what is actually working?

The honest AI conversations right now are happening in small, closed, peer-led rooms — not on the main stages of the industry's biggest events. And here is the thing: this is not a people problem. It is a math problem.

I have sat through enough AI sessions at major conferences to recognize the pattern within the first ninety seconds. A vendor walks on with a polished deck. A consulting firm presents a framework so broad it could apply to any organization in any sector — which is precisely why it is useless for yours. The frustration behind the question — where do I go to hear what is actually working? — is legitimate, and more widespread than conference organizers want to admit.

Why do most AI conference sessions feel like sales pitches?

Because most of them are — and it is worth being fair about why. A seat on a big conference stage costs five figures. Once someone has paid that, the talk has to earn it back, so the content bends toward the pitch. Not because anyone is a villain; the economics leave no other option. The vendor is a hostage to the cost structure as much as you are. Blame the model, not the person standing in front of it.

That math has a cost you feel in the room: the signal-to-noise ratio drops low enough to exhaust even a patient leader. A widely cited MIT study — The GenAI Divide, from MIT's NANDA initiative — found that 95 percent of enterprise generative AI pilots deliver little to no measurable financial impact. Yet the stage still tells you adoption mostly works. The people actually running deployments will tell you, quietly and off to the side, how often it does not. That gap between the stage story and the operational reality is the fatigue you are feeling.

What works better than a bigger conference?

A different structure works better — not a different venue. The events worth your time are built around a specific problem instead of around who paid to present. Three things change when the structure is right.

Why does a small room matter so much for AI conversations?

Small matters because of what it mechanically buys you — not because “intimate” sounds nice. Small and low-cost means no one in the room needs to pitch to justify the seat, so the information stays honest. Small means you can actually hear each other. Small means the person across from you will say “that did not work for us,” because it is not going to end up as a LinkedIn post. Every one of those outcomes traces straight back to the format. The size is the mechanism; the candor is the result.

There is a finding underneath all of this worth sitting with — and it comes from the same MIT research: when AI deployments fail, the cause is almost never the model. It is the learning gap between the tool and the organization — data quality no one mapped, governance no one owned, workflow assumptions that broke on contact with reality. That kind of specificity does not survive a main stage. It lives in peer conversation.

What should a leader check before registering for an AI event?

A five-point filter, before you give anyone your badge scan:

1. Cross-check the speaker list against the sponsor list. Heavy overlap means the content is likely promotional — again, not malice, just the math.
2. Look for post-mortems, not just success stories. Someone willing to walk through a failure publicly has nothing to sell you.
3. Favor small-table formats over ballroom keynotes. Depth per hour beats attendance count every time.
4. Ask whether there is a genuine no-pitch policy — and whether it is enforced or decorative.
5. Prefer repeat cohorts over one-off crowds. Trust compounds; the same peers quarter after quarter will teach you more than three conferences of rotating strangers.

Where does In Good Company fit into this?

In Good Company is exactly why this argument is not theoretical for us — we built it on these principles before writing about them. Small rooms, no pitches, built around one real problem at a time, with people in the seats who have actually solved it. Not because we are the AI oracle. We are not. It is because the room is smarter than any stage, and our job is just to build the conditions where that shows up.

It is also why, this August, MTMG and Dirigo are launching Leadership in the Age of AI — a conversation series built on the same premise: practitioners talking about what is actually working, with nothing for sale. We would rather build the room than complain about the ballroom.