# How General Intelligence used agents to build an agent platform on Vercel

DevFeed: [How General Intelligence used agents to build an agent platform on Vercel](<https://devfeed.tech/articles/how-general-intelligence-used-agents-to-build-an-agent-platform-on-vercel-741.md>)

Original publisher: [Read original article](<https://vercel.com/blog/how-general-intelligence-used-agents-to-build-an-agent-platform-on-vercel>)

Author: Eric Dodds

Published: 2026-05-04T04:00:00Z

Content type: article

Language: en

Sources: [Vercel News](<https://devfeed.tech/sources/vercel-news.md>)

Topics: [AI Bots](<https://devfeed.tech/topics/ai-bots.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [apis](<https://devfeed.tech/tags/apis.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [sre](<https://devfeed.tech/tags/sre.md>), [vercel](<https://devfeed.tech/tags/vercel.md>)

## AI overview

General Intelligence describes using Vercel's programmatic cloud controls to support Cofounder, an AI-agent platform. The article argues that coding agents need API and CLI access to infrastructure operations such as deployments, DNS, billing, configuration, and logs.

## Source excerpt

General Intelligence on Vercel 8-person team (5 engineers) shipping 10 PRs and 70+ commits per engineer, per day 4,000+ preview branches with ~100 parallel app versions running at any moment 90% of SRE work automated through Vercel and their own agent (Cofounder) Cofounder launches with a managed Vercel account for every customer General Intelligence is building a platform that lets any founder run a company entirely with AI agents. Their vision is the one-person, billion-dollar company where every department is driven by agents. Their flagship product, Cofounder, gives founders a full team of agents that cover engineering, marketing, SEO, finance, sales, customer support, and operations. General Intelligence is an 8-person company with 5 engineers. To ship a platform that lets their customers run agentic companies, they had to operate as one themselves. They wanted to use Cofounder's CTO agent to build out other agentic business functions, and their developers quickly learned that was an infrastructure problem, not an agent problem. For a complex, multi-tenant platform, the requirement was total programmatic control: every action a human could take on the underlying cloud platform had to be available to a coding agent through a CLI or APIs. That meant killing deployments, changing DNS, managing billing, editing configs, all of it. Most cloud providers failed that test, and that's why General Intelligence migrated to Vercel. Agents need infrastructure built for agents A coding agent runs dozens of processes in parallel. It queries logs as data, parses errors as input, and treats every dashboard click as a missing API. The bottleneck stops being how fast the agent can write code and becomes how much of the cloud it can actually reach. Most clouds aren't shaped for that pattern. They're built for human developers: dashboards to click, consoles to read, APIs that cover some operations but stop short of others. The closer an agent gets to running development end-to-end,