The conference badge said "Navan Chauhan, StrongDM" and below it, in smaller type, "Speaker." He'd worn one like it before, at a smaller venue, months earlier. That talk had been about the Digital Twin Universe. The audience had listened politely. The Q&A had been dominated by a single question, asked in various forms by various people: "So these are just mocks?"
No, they were not just mocks. He had explained this. Mocks returned canned responses. Digital twins replicated behavior. Mocks tested your code against your assumptions. Twins tested your code against reality. The audience had nodded and moved on and probably gone back to writing mocks.
That was then.
This time the conference was larger. Two thousand attendees. The talk was in the main hall, not the breakout room. Navan stood at the podium with the same slides—updated, refined, but structurally the same—and gave the same talk. Behavioral clones. API surface fidelity. Edge case reproduction. Scenario-driven validation. The satisfaction metric.
Nobody asked about mocks.
The first question from the audience was: "How do we build our own twins?"
Navan paused. Not because he didn't know the answer—he had slides for this—but because the question itself was the answer. The world had caught up. Between his first talk and his second, something had shifted. People had read the StrongDM paper. They'd seen the Stanford Law CodeX write-up. They'd read Simon Willison's analysis. They'd tried building AI-driven development pipelines and discovered that mocks weren't enough, that you needed something with higher fidelity, something that behaved the way the real world behaved.
They needed twins.
"You start with traffic captures," Navan said, advancing to his next slide. "Real API interactions. You instrument the real service and record every request and response. Then you build a behavioral model that reproduces those interactions. Not a replay—a model. It has to generalize. It has to handle requests it's never seen before, the way the real service would handle them."
The questions kept coming. How do you validate twin fidelity? How do you handle API version drift? What's the latency overhead? Can the twins scale beyond single-node?
Real questions. Implementation questions. The kind of questions people ask when they've already decided to build the thing and need to know how.
After the talk, a line formed at the speaker's table. Navan talked to engineers from twelve different companies. Three of them were already building digital twins for their own services. One had read the CXDB source code on GitHub. Two asked about Leash. One asked if StrongDM was hiring.
Navan texted Justin from the hotel lobby that evening. Nobody asked about mocks.
Justin's reply came in thirty seconds. The world caught up. Now stay ahead.
Navan opened notebook #8 to a fresh page and began sketching the next generation of the twin architecture. The world had caught up. That meant the factory needed to be somewhere else entirely.
He wrote until the hotel bar closed. The bartender had to ask him twice to leave. He didn't notice the first time. He was drawing diagrams.
"Nobody asked about mocks." Four words that tell you more about industry adoption than any market research report. The world caught up and Navan was there to see it happen.