Interview · Engineering“We rewrote our sync engine twice—and it was worth it”
Acquired FM18 min listen
When the team hit 40,000 active syncs a day, the original architecture started showing its age. The second rewrite cut p95 latency nearly in half, but the real win was simplifying how conflicts resolve.
“Most users never think about sync conflicts, and that is the point,” the co-founder explains. “The moment they do, we have failed.”
For the founding team, the lesson was simple: their earliest assumptions about how people collaborate were wrong in ways that only surfaced at scale.