There's a video you should watch before reading this. It's three minutes long and it explains more about movements than any business book.

A guy starts dancing alone on a hill at a music festival. He looks ridiculous. Nobody joins. He keeps going. Then one person joins — and now there are two idiots dancing. Then a third. And somewhere between the third and the fifth, it stops being embarrassing and starts being a movement. Within a minute, hundreds of people are running to join.

The lesson: the leader didn't start the movement. The first follower did. The leader was just a lone weirdo until someone validated them by joining.

Search "first follower leadership lesson" on YouTube. Watch it. Then come back.

B1C3 Is the Guy Dancing Alone

One person. One month. 35 repositories. A manifesto. A patent. Outreach to UNESCO. Code pushed to the wrong GitHub account that turned out to be the right one.

Nobody asked for any of it. Nobody approved it. Nobody was watching.

That was the first move. It's been made. Now the question is: who's second?

What You Can Actually Do

This isn't "join our vision." This is: here's real work, right now, that needs a human.

You Do CAD or 3D Scanning

We're building intelligent point cloud compression — a tool that takes 30 million scanned dots and asks "how many do I actually need?" A flat wall is 50,000 dots. It should be 4.

Your domain expertise is the product. Algorithms detect shapes. You know what those shapes ARE — a DN150 pipe, a 90-degree elbow, a flange. That knowledge is the moat no algorithm replaces.

What you'd do: Validate surface classifications. Label training data. Provide real scan data. Shape the product from the domain side.

You Do ML / AI

We need surface classification models trained on real point cloud data. The pipeline is documented — RANSAC for baseline, PointNet++ for the ML layer, Open3D and CGAL for processing.

Public datasets exist (ScanNet, S3DIS). Domain-specific labeled data doesn't — we're building it.

What you'd do: Train models. Build the classification pipeline. Evaluate against traditional approaches. Push the accuracy where algorithms alone can't reach.

You Do Frontend

We need a 3D point cloud viewer — interactive, web-based, showing original vs. compressed side by side. Color-coded surfaces. Compression stats. Runs on a laptop and a tablet.

What you'd do: Three.js or Deck.gl. WebGL rendering of large point sets. Dashboard for compression metrics. Make the invisible visible.

You Do UX

Our tools need to work for the 80% — people whose brains work differently. Current CAD and scanning tools are built for one cognitive profile. We're building for all of them.

What you'd do: User research with real surveyors and engineers. Design the upload → process → download flow. Accessibility testing. WCAG 2.1 AA. Make complexity feel simple.