A small search win: Google's AI Overview cited my Figma multiplayer teardown

I write about production system design on Buildopsy. Recently, I published a teardown of Figma’s multiplayer architecture: how edits move between browsers and servers, how conflicts get resolved, and where the system’s scaling costs come from.

I later searched for “how Figma’s multiplayer technology works.” Figma’s own engineering post appeared in the results. In the AI Overview below it, I noticed two Buildopsy citations beside explanations of optimistic rendering and the server’s role in ordering edits.

What appeared in the result

I saved a screenshot because I didn’t expect to see my article cited there. It’s one result for one query—not proof of a lasting ranking or a traffic increase.

The interesting part is how edits get reconciled

A browser applies an edit immediately, then sends the operation over a WebSocket to the process that owns that document. The owner orders the change and broadcasts the accepted version to the other clients.

The conflict rule is more specific than “the server wins.” In the architecture described in my teardown, the server orders operations, same-property conflicts use last-writer-wins, and tree changes get an additional validation step to prevent cycles. That’s a product-specific trade-off: a design canvas can tolerate some conflicts that would be painful in a collaborative text editor.

The Rust rewrite addressed another boundary. A slow document could stall unrelated documents sharing a TypeScript worker. Figma moved document operations into a Rust child process while Node.js kept handling the network layer. Figma reported that serialization became more than 10× faster in its engineering write-up.

The numbers make the next question more interesting. In one scenario, 1.5 million concurrent users sending two coalesced cursor frames per second produce 3 million ephemeral frames per second. That’s modeled math, not a Figma disclosure. The design question is: which events need durable ordering, and which can be dropped when they’re stale?

The full teardown works through that boundary, a 65k-RPS scenario, and a modeled monthly cost range. The cost figures are my assumptions, not Figma’s reported spend: Read the Figma multiplayer teardown.

I don’t know whether the citation will persist or bring readers. It was still encouraging to see a technical explanation appear alongside the primary source it discusses.