Putting the Patterns to Work

*clip from this month’s live IOL session with our guest, Leo Chan, LEGO® SERIOUS PLAY®

Last month I shared the founding story of Regenerous Labs — how a moment of clarity in a chemo chair led to a two-year exploration with trusted collaborators, and an architecture nobody predicted. Start With Who, practiced.

Since then, the weekly Innovating Out Loud series has shifted. For the first three months of 2026, I was exploring ideas — building vocabulary, testing frameworks against research, finding the connective tissue between the book’s four patterns and what I’m learning as a venture builder. That exploration produced concepts like dormancy, vital signs, complementary constraint, and the factory-versus-forest metaphor for how we organize AI.

Over the past month, the work changed. I stopped writing about innovation and started showing it. Three experiments went public. Each one tested the book’s frameworks in the real world. And each one taught me something the frameworks alone couldn’t.

[Links to all three below]

Experiment 1: Innovation Coach

I built a diagnostic tool with the 77 innovation frameworks from The Insider’s Guide embedded in its architecture, powered by a stripped-down version of our behavioral intelligence engine. I ran the same problem — a real recruiting challenge the Lab is facing — through Innovation Coach and through several leading LLMs.

The LLM output was genuinely useful. Competent, structured, actionable. A smart strategy.

Innovation Coach did something different. It didn’t answer the question I asked. It questioned my question — then answered with an analysis that addressed what I actually needed.

The difference isn’t the knowledge. It’s the perception. Innovation Coach runs on the same underlying AI models. What’s different is the architecture: multiple expert lenses at different apertures running in a diagnostic sequence before looking for answers. Diverge-Converge-Synthesize — the rhythm from the book — operationalized in software.

Readers of the book will recognize what I was doing: developing in the open, the practice from the VS Code story. What gets seen gets tested. 150 people tried it. The feedback was honest: the reframe was something no general LLM produced. But the actionable output needed work.

Fair. So I built the next one.

Experiment 2: CANOPY

CANOPY is a four-step decision intelligence journey built in a single day using Claude as my full IDE — chat to design to code to deployment. Multiple expert lenses run independently, then converge. The tool reframes your question and coaches you through creating a commissioning brief and a 90-day action plan.

The architecture held. The pipeline worked.

But here’s what I lost.

On the Friday before the build, I sat in a design session with my partners Dan and Kent for one of the Lab’s expert products. What I built alone on Saturday over eight hours isn’t in the same league as what they produced in ninety minutes — and you wouldn’t need an expert eye to see it.

Worse: privacy by design, transparent AI, confidence scoring — requirements I know matter — fell out of my head entirely while I was designing the core workflow. I bolted some back on at the end. But bolting things on after the fact undermines the whole discipline of Diverge-Converge-Synthesize. I know that better than anyone. Building alone, even with excellent tools, I could only actively hold so much.

The lesson underneath the lesson: the expert building alone, even with AI that extends her capability, loses signal that the expert building with complementary collaborators wouldn’t.

That’s not a failure of the tools. That’s Pattern #3 — Innovating With Everyone — proved from the inside.

Experiment 3: The Data Center Blueprint

For three years I’ve had a technical and commercial blueprint for a community-scale regenerative data center. Modular, flatbed-deliverable, deployable in 90 to 120 days. AI-class compute with waste heat routed into greenhouses and district heating. Grid-interactive, community-embedded, and more economically viable over a ten-year horizon than hyperscale warehouses.

Every component is commercially available. Microsoft’s own Fairwater architecture — connecting the Wisconsin and Atlanta campuses via 120,000 miles of dedicated fiber as a single distributed supercomputer — proved the networking layer works at scale. An April 2026 study showed 3 to 21 percent cost reduction in grid investment from geographically shiftable AI workloads. FERC is advancing rules to expedite interconnection for flexible loads.

Everyone I walk through the blueprint says yes, obviously. Then nothing happens.

Four actors are each defending structures the current system rewards. Utilities defending a capex-driven rate base. Developers defending a construction-company operating model. Communities defending themselves from extraction — reasonably, given that $156 billion in data center projects were blocked by local opposition in 2025. Capital defending a hyperscale thesis committed in 2023.

Each one is rational. Each one reinforces the others. None unlocks alone. The technology isn’t the barrier. Behavior is the barrier. The line Dean and I wrote about 70 percent of transformations — it holds at infrastructure scale.

I published the full blueprint as prior art and coalition invitation. Because hiding the specification is the surest way to guarantee it stays unbuilt. The behavioral lock that’s kept the obvious solution unbuilt for three years is the same lock that protects it from capture. No utility, developer, or hyperscaler will take it to market — because their internal incentive structure forbids it.

The lock is the moat. And the coalition needs more minds than mine.

Where the Patterns Stand

Three experiments. Four patterns tested.

Pattern #1 — Innovating Every Day. The weekly IOL practice itself. Seventeen pieces. Saying it ugly, sharing before it’s perfect, learning in public. The vocabulary has grown from five terms to more than thirty — none existed in the book, all grew from the book’s root system.

Pattern #2 — Innovating Over the Years. We’re fourteen months into the adaptive cycle. Seedling is becoming young forest. The tools are in the wild. The vocabulary is accumulating. The experiments are teaching us things we couldn’t learn any other way.

Pattern #3 — Innovating With Everyone. CANOPY proved it by accident. The expert building alone drops things the team wouldn’t. The architecture holds. The principles need more eyes.

Pattern #4 — Innovating More Than Technology. The data center piece is the sharpest case. Every technical component exists. The missing innovation is the integrated system, the governance model, and the coalition.

The patterns hold. They’re generating new tools. And the community that uses them and pushes back on them is what makes them sharper.

That’s you.

Try the Tools

Both are experiments, not products. Your data stays local. We can’t see what you do.

Innovation Coach — Give it a real problem. See what it reframes that a general LLM doesn’t. www.regenerouslabs.com/innovationcoach

CANOPY — Walk through a four-step decision intelligence journey with multiple expert lenses. A preloaded test case lets you experience it without bringing your own. www.regenerouslabs.com/canopy

Tell me what I got wrong. Help me build them better.

Go deeper:

📖 The Insider’s Guide to Innovation at Microsoft

🎓 LinkedIn Learning course

📬 Innovating Out Loud weekly series — where all of this lives between monthly updates here, including our live webcast every 4th Thursday of the Month

🎤 Speaking and workshops — Dean and I bring the patterns to leadership teams and conferences. For internal events at Microsoft, reach out to Dean on Teams!

AI Disclosure: This piece was created with the help of AI — specifically Claude and a team of expert personas built by Regenerous Labs. Direction, judgment, and final decisions by me. Say it ugly, build it better. Onward!