One Next.js Backend, One Billion Requests a Day

Factory builds an AI platform that automates parts of the software development lifecycle, from signal intake to production deployment. The company's positioning is not about replacing engineering judgment, but about eliminating the repetitive work around it. Internally, Factory applies the same philosophy to its own operations, remaining lean by deploying its "Droids" to handle routine workloads.

As Factory grew from a simple web app into a multi-surface platform serving enterprises with tens of thousands of engineers, its backend scaled with it. A single production Next.js app on Vercel now powers all surfaces, handling one billion daily requests across API routes, middleware, and webhook handlers. This scale has not required turning backend maintenance into a dedicated engineering project.

"We're doing tens of millions of API requests daily. We've been on Vercel since the beginning and infrastructure has never slowed us down."

— Nikita Lilichenko Technical Staff @ Factory

Treating Next.js as a Full-Stack Cloud Backend

While most teams treat Next.js as a frontend framework, Factory runs its entire cloud backend on it. The customer-facing API runs in its own section of the Next.js app, middleware handles authentication and routing across all surfaces, and webhook handlers plus log drain pipelines feed analytics downstream. As the platform grew, new workloads were absorbed into the same backend without standing up separate infrastructure or introducing new vendors.

The only parameters requiring tuning were memory limits and function duration, both of which scaled up without incident. Vercel's Fluid compute keeps functions warm between requests, which removes the cold start penalty typically associated with serverless and keeps the p95 response time at or below 350ms. After three years, growth has come without the operational overhead usually attached to it.

"When someone needs to write a new route that our web platform needs to use, it's a no-brainer. We just spin it up in Vercel."

— Nikita Lilichenko Technical Staff @ Factory

Letting Non-Technical Teams Deploy Their Own Tools

Factory's non-technical employees once had to route every internal request through engineering, whether it was an account executive needing an analytics dashboard or a solution architect building a customer prototype. Each context switch cost the engineering team shipping time. Factory turned its own product on the problem.

Droids in the Factory Desktop App or CLI now handle this work. Non-engineers use them to build and deploy internal tools directly to Vercel, with Droids hitting Vercel's APIs to manage configuration, builds, and deployments end-to-end. Engineers set guardrails and review high-stakes decisions, while Droids manage the rest. This yields dozens of deployments per day across the team and a growing library of internal assets, including:

  • Internal analytics dashboards pulling from Factory's analytics databases and software tools
  • Customer-specific demos that solutions engineers can build and deploy in minutes
  • Custom utilities for one-off internal workflows

Securing the Backend Against Automated Abuse

Opening self-serve signups introduced a new threat category. Bots created fraudulent accounts and requests flooded in from bad IP addresses, requiring route-level protection at the network layer. Vercel Observability provides visibility into what is hitting the API, surfacing traffic patterns and problematic IPs as they appear. Enforcement is handled by the Web Application Firewall, which applies route-level rules to block bad actors while rate limiting and DDoS protection operate continuously underneath.

"We're doing tens of millions of requests daily and I'm not losing sleep over what's hitting the API. Vercel's WAF and rate limiting are just always on, always working."

— Nikita Lilichenko Technical Staff @ Factory

Dogfooding at Scale

Factory uses its own Droids to build Factory. This is both an operational model and a testing ground: internal automations become product features stress-tested on the company's own infrastructure before being shipped. Factory was founded in 2023 by Matan Grinberg and Eno Reyes, is headquartered in San Francisco, and is actively expanding its platform toward autonomous software factories at enterprise scale.

"We're growing rapidly and trusting Vercel to scale with us."

— Nikita Lilichenko Techncial Staff @ Factory