Factory, a startup using artificial intelligence agents to speed up software projects, announced on Tuesday that it has raised $200 million in funding to scale its self-improving software development platform across the enterprise market. The round was backed by Blackstone, Khosla Ventures, Sequoia Capital, NEA and several other prominent investment firms, joined by a group of angel investors that included Marc Benioff. The raise values the company at $5 billion, a $3.5 billion increase from its April valuation, according to announcement coverage. Factory, officially The San Francisco AI Factory Inc., offers a platform that can automate complex programming tasks such as porting an application to a new language. Developers interact with the software by entering high-level descriptions of the tasks they wish to perform, and Factory breaks down each task into sub-steps and assigns them to a group of AI agents. "Across the world's largest enterprises, we are seeing a move from individual coding agents to software factories that serve as the core foundation from which an entire software company operates," said Factory co-founder and Chief Executive Officer Matan Grinberg. ## How the Platform Runs a Software Project Under the hood, the platform automatically provisions infrastructure for each agent and performs related tasks such as preventing agents from interfering with one another's work, while a monitoring dashboard tracks project progress in real time. The starting point of Factory-powered projects is a slash command called Readiness Report, which instructs the platform to check whether the code repository being edited is agent-friendly. AI agents struggle to process code files that are not well-organized or lack technical explanations. Factory also checks whether the repository features guardrails that prevent agents from introducing vulnerabilities. If everything is in order, the platform begins the coding task, assigning each step to the large language model that can perform it in the most cost-efficient manner. If an LLM experiences technical issues, the platform reroutes the request to a different algorithm. Once a code snippet is ready, Factory runs it through a series of quality and cybersecurity checks. A feature called AutoWiki generates documentation that explains the code and automatically refreshes it after every update. Developers can also configure an AI agent to monitor code for technical issues and generate troubleshooting suggestions. ## Droid Computers and Self-Improving Agents Factory runs its agents in virtual environments called Droid Computers, which customers can deploy in the cloud or on their own infrastructure. A Droid Computer saves the software libraries, configuration files and other technical assets used by the agents it hosts, removing the need for developers to manually install those assets before every session. In January, the company detailed that it is using LLMs to automatically improve its platform's AI agents, with the models making enhancements based on anonymized usage signals collected from thousands of daily sessions. That self-improvement loop is central to the company's pitch that its platform compounds in value as it is used, learning from thousands of daily enterprise sessions. The company disclosed alongside the funding that its platform is used by hundreds of thousands of developers, some of whom work at major technology firms such as Nvidia Corp., Adobe Inc. and Palo Alto Networks Inc. ## Headcount to 300 as Enterprise Demand Grows Factory will reportedly use its new funding to grow its headcount to 300 employees by the end of the year, scaling sales and engineering as it chases the enterprise shift Grinberg describes, nearly tripling the organisation behind one of the most closely watched agentic coding platforms. The raise lands in a market that has moved quickly from experimentation with individual AI coding assistants toward platform-level adoption, where entire development pipelines are orchestrated by fleets of specialized agents. Factory's approach of wrapping agents in governance, cost routing, security checks and automatic documentation is aimed directly at the concerns that have slowed enterprise deployments. With backing from some of the most prominent names in technology investing and a valuation that has nearly quadrupled in six months, the company now faces the task of proving that agent-built software can meet the reliability bar of the world's largest engineering organizations, a test its growing roster of marquee customers suggests is already under way. The economics of the model are also part of the pitch. By routing each sub-task to the language model that can complete it most cheaply and rerouting around outages, Factory is betting that cost efficiency at scale will matter as much to enterprise buyers as raw capability, an argument its investors appear to have accepted with a valuation that has climbed steeply in half a year. For developers, the platform's promise is that entire tasks, from preparing a repository through generating documentation, can run as one supervised pipeline rather than a sequence of manual prompts, a shift that reframes the engineer's role from writing every line to directing and reviewing the output of agent fleets.