United We Transform began as a live experiment: a government grant, the world's first AI workspace, and a one-day gathering of more than a thousand people built to test how humans and AI collaborate better together. Some of it worked brilliantly. Some of it did not. All of it is why this atlas exists.
Direct answer: United We Transform built the Agenda Intelligence Atlas after running its own large AI-era gathering, learning that even ambitious event design can fail without clear follow-through, participation architecture, and evidence.
Inside the 2024 gathering that started United We Transform. Plays muted; tap the video to turn on sound.
United We Transform started with a government grant to explore a simple, stubborn question: in the AI era, can we make the way people gather and work together dramatically more effective?
We opened the Exchange in Saint Paul (370 Wabasha Street North): movable walls, writable surfaces, no screens dominating the room, and discreet AI quietly capturing insights in the background so people could stay face to face.
Inside the Exchange we developed the UWT method: every hour the room Scans (who is here, what changed), Focuses (the core issues), and Acts (moves on decisions immediately), with AI check-ins keeping everyone aligned. The goal: compress months of work into hours.
We gathered roughly a thousand participants (1,245 registered) and ran fifteen live experiments at once: no registration table, no rows of chairs, no fixed agenda, reinvented panels, AI-matched networking, and more. A stress test of everything we believed about gatherings.
7,545 networking interactions were recorded, the most outcome data we can find a record of at any event. We also published what did not work: only 48% of AI networking recommendations were rated positive, our tooling buckled at 500 simultaneous users, and up to 40% of the room was confused at times. The full deck stays live on our learnings site.
Running our own experiments taught us how rare real outcome design is, and how hard follow-through is even when you are trying. So we turned the lens outward and scored 23,624 public event agendas. That is this site: the Agenda Intelligence Atlas, plus the Exchanges that now operate as permanent physical spaces.
The 2024 experiment taught us that pairing people with AI in a room does not happen automatically. Somebody has to design the room on purpose, then check afterward whether it actually worked. Brandon Klein later gave that role a name: the Cognitor, the person who orchestrates human judgment and AI capability together instead of leaving the room to run on autopilot.
He introduced the term first, in a LinkedIn article in May 2025. A month later he and Mark Rowland published it properly, in "Smarter Together," for Duke Corporate Education Insights, June 2025.
This atlas is a Cognitor's instrument: something a person uses to design a gathering with AI, then track what changed because of it. Read the full concept on the Cognitor page.
We publish our misses next to our wins, because that is the standard we hold every gathering to. Eight of the fifteen, with the honest results:
The other seven, detailed in the full learnings deck:
All fifteen experiments, the data, and dozens more learnings from the 2024 gathering.
We learned firsthand that designing a gathering for outcomes is brutally hard: even with AI in the room and every convention questioned, follow-through was our weakest result. When we later scored 23,624 public agendas, we found the whole field shares that gap: follow-through averages 5.8 out of 100. The atlas, the Gathering Effectiveness Score, and the free grader are the tools we wish we had before May 21, 2024.
The work moved from a single room to a public benchmark. Instead of telling event teams to copy our format, we built a way to read any public agenda for visible mechanisms: participant work, learning transfer, network design, commitments, tracking, and evidence. The history matters because the atlas is not abstract commentary; it came from trying, measuring, missing, and then making the standard public.
Because the same trust rule applies to us: useful gathering design needs visible evidence, not just a polished story.
The 2024 experiment made follow-through and outcome evidence painfully concrete, which shaped the eight-pillar Gathering Effectiveness Score used across the atlas.