A CEO shut down his entire company for one week and told everyone to build with AI. The results are all over your feed right now. The reason it worked is not in the clip, and if you copy the week without understanding the reason, you will pay for the demo and skip the outcome.
Here is what actually happened.
What Boom Supersonic actually did
Blake Scholl runs Boom Supersonic, the Denver company building the first new commercial supersonic airliner since Concorde. Last week he sat down with Reid Hoffman on the Possible podcast and told the story that's now circulating as a 60-second clip: Boom paused all project work for one week and told every employee to build something with AI.
The design was specific. Small teams. Each team picked whatever they judged to be the most important thing they could see for the company. AI use was mandatory. And at the end of the week, every team demoed in front of the entire company.
Scholl expected two or three creative things that would justify shutting the company down for a week, and dozens of duds.
He got the inverse. Two or three duds. Dozens of genuinely good projects. And two or three that he says changed the company's trajectory.
The headline example is a tool called Blade Runner. Two people built it. One is an engineer. The other is a former engine engineer who disliked the big engine companies enough to quit and become a software engineer. The tool puts three panels on one screen: blade geometry you can adjust with sliders, a real-time structural simulation, and a real-time aerodynamic simulation running on GPUs. An engineer drags a slider and watches the physics respond in moments. There are roughly a thousand blades in a jet engine. Blade Runner means a couple of people can now design and optimize all of them.
At the other end of the org chart, the person who runs shipping and receiving built a shipping and receiving tool.
Scholl's conclusion: everybody actually has an idea. The week proved it.
It's a great story. It's also being read almost entirely wrong.
The most important sentence in the interview is the one nobody clips
Before the AI week existed, Scholl tried the thing most executives try. He stood up at all-hands meetings and told everyone to use AI.
His words: nothing was happening. Or very little was happening.
Sit with that for a second. A founder with a software engineering background, running a company dense with technical talent, personally telling his people from the stage to adopt AI. And it did not work.
The mandate failed. The structure worked.
This is not an AI story. It is an adoption story. The tools were available before the week and after the week. What changed was the design: protected time, small teams, ownership of the problem, a mandatory public demo. Every one of those elements is a human on-ramp, not a technology.
I have watched this exact pattern play out in my own client work. An insurance agency I worked with had paid Claude and ChatGPT subscriptions sitting unused for months. The owner had told his staff not to touch AI until the company formally rolled it out. Deployment complete, adoption zero. Then we ran one structured two-hour session with clear guardrails and a real workflow, and by the end of it his operations manager, a self-described AI skeptic, said out loud: "It's clicking."
Same tools. Same people. The difference was the on-ramp.
That is what Scholl built. A very good on-ramp. The clip sells it as a lightning strike. It was engineering.
What this story does not mean
Here is where I want to push back on the takeaway spreading through executive feeds right now, because I can already see the calendar invites going out.
This story does not mean you should schedule an AI hackathon next month.
The executive watching this clip and booking a company-wide AI week as the intervention is copying the spark and skipping the tinder. And Boom had years of tinder.
Consider what already existed at Boom before the week started. The founder is a former Amazon software engineer who built the company's culture around software from day one. Boom had already built its own airplane design software, before AI coding tools existed. They embed software engineers directly inside hardware teams. They built their own CAD kernel, which is the piece of software Blade Runner's real-time geometry actually runs on. They had already built an AI retrieval system on top of their own regulatory documentation.
The AI week did not create builders. It gave existing builders permission and a deadline.
Now look at the two details in the story that quietly confirm this. Blade Runner was not finished in the week. Scholl says it took two people something like three weeks, working with an AI coding tool. The week surfaced it. The foundation finished it. And the bar every project had to clear was a demo in front of the company. A demo is not a deployment.
The distance between a demo and a department is where the real money lives. A tool that one person runs on their own machine, on their own data, when they remember to run it, costs almost nothing. The same tool serving five hundred people every day has a token and API bill that is a real line item. It needs an owner when the model changes or the builder leaves. It needs a security review, because a personal experiment touches one inbox and a department tool touches everyone's. None of that appears in the interview, and I don't fault Scholl for it. At a company that dense with engineers, that overhead gets absorbed by the culture. At most organizations, it does not get absorbed. It gets discovered, later, as an invoice.
So no, the lesson is not "shut down for a week and magic happens." A few hundred people at an aerospace company with a software founder is not your org chart.
My read: the week was the last step, not the first
Everything in this story confirms something I tell leadership teams constantly, and it's the part that doesn't clip well because it isn't exciting.
AI amplifies what is already there.
Boom's week produced trajectory-changing tools because Boom had spent years building the thing the AI could amplify: documented ways of working, software culture, people who understood their own workflows cold. The former engine engineer who co-built Blade Runner did not learn blade design during the AI week. She carried a decade of clarity into it, and the AI turned her clarity into software.
Run the counterfactual. Take the same week, the same mandate, the same demo-day pressure, and drop it into an organization where workflows live in people's heads, where nobody has documented how work actually moves, where there is no AI policy and no shared understanding of what data can touch which tools. You will still get a demo day. You will get slick prototypes built on unapproved accounts with data nobody cleared, owned by no one, priced by no one. Some of them will even look like Blade Runner for about a month.
Chaos in, chaos out, faster.
The dividing line is not bold companies versus timid ones. It is companies that built the foundation versus companies that bought a week of theater.
What to do with this story now
If the Boom clip crossed your feed and something in you said "we should do that," good. That instinct is correct. The sequencing most leaders will follow is not.
Do not start with the week. Start with what Boom had before the week.
First, get your executive team literate enough to evaluate what a demo day would actually produce. Not expert. Literate. If your leadership cannot tell the difference between a prototype and a production tool, demo day will mislead you at the exact moment you feel most informed.
Second, put governance in writing before anyone builds. Which tools are approved. What data can and cannot go into them. Who reviews what gets built. This is a document, not a program. It takes hours, not quarters, and it is the difference between an innovation week and a security incident with a highlight reel.
Third, have your teams document the workflows they'd want to improve. The Blade Runner builders knew exactly which thousand-blade problem they were attacking. Clarity in, clarity out.
Then run the week. Small teams, self-selected problems, mandatory demo. Scholl's design is genuinely good, and by that point your version of it will surface real builders and real problems instead of theater. And when the demos land, ask the two questions the clip never will: what does this cost at full headcount, and who owns it when it breaks.
The week Boom shut down is the story everyone is sharing. The years Boom spent getting ready for that week is the story worth copying.
The clip shows you the spark. Your job is the tinder.
