Meta's new Superintelligence Labs executive is pushing staff to ditch slow internal systems for faster engineering tools (www.businessinsider.com)

Meta’s newly formed Meta Superintelligence Labs (MSL) is pushing engineers to ditch slow, homegrown deployment tooling in favor of mainstream platforms — notably Vercel paired with GitHub — to speed prototyping and iteration, according to internal memos obtained by Business Insider. Leaders including Nat Friedman (head of PAR) and chief AI officer Alexandr Wang are steering teams toward these external tools because Meta’s infrastructure, built for billions of users, imposes hours‑long deploy times and “is not conducive to vibe coding.” MSL is already running about 10 projects on Vercel/GitHub and aims to cut deployment latency from roughly 99 minutes to under two minutes; it’s also building an internal “Nest” platform (TypeScript-focused) as a longer‑term option. Friedman and Wang are investors in Vercel, which has raised significant funding and counts large customers like Netflix and Adobe. This move matters because faster developer feedback loops are critical for rapid AI experimentation and productization amid Meta’s multimillion‑ to nine‑figure AI hiring bet and competition with OpenAI and Google. Technically, the shift highlights tradeoffs between scale-optimized internal infra and nimble, cloud-native developer experiences: external platforms can reduce iteration time and enable “vibe coding” (AI-assisted coding), but raise integration, data‑hosting and security questions that MSL says Vercel is willing to help mitigate. The dual strategy — short‑term unblock via Vercel plus a bespoke Nest for long‑term integration — signals Meta’s pragmatic pivot to speed while it retools its stack for next‑gen AI workflows.