OpenAI CEO Sam Altman unveiled AgentKit, a new toolkit designed to help developers build autonomous AI agents, during the company’s DevDay 2025 conference in San Francisco. The launch marks OpenAI’s biggest developer-focused initiative since ChatGPT and signals a major move into the fast-growing autonomous agent space.
Described by Altman as “a complete set of building blocks to take agents from prototype to production,” AgentKit is built to simplify the creation of advanced AI systems capable of managing complex workflows. The toolkit comprises four main components: Agent Builder, a drag-and-drop visual interface likened to “Canva for agents”; ChatKit, an embeddable chat interface for applications; Evals for Agents, offering testing and performance analytics; and access to OpenAI’s connector registry for secure third-party integrations.
In a live demonstration, OpenAI engineer Christina Huang created two functional AI agents in under eight minutes, showcasing the platform’s accessibility. “This is everything we wished we had when building our first agents,” Altman said, highlighting OpenAI’s goal to make agent development faster and more intuitive.
Altman also revealed that ChatGPT now boasts 800 million weekly active users, up from 700 million just two months earlier. The milestone comes alongside the rollout of OpenAI’s Apps SDK, which lets developers embed interactive services from platforms such as Spotify, Figma, Coursera, Zillow, and Canva directly into ChatGPT conversations.
In a further expansion of its ecosystem, OpenAI announced a multi-billion-dollar partnership with AMD to deploy up to 6 gigawatts of GPUs over several years. The agreement includes an initial rollout of AMD’s Instinct MI450 GPUs in late 2026 and gives OpenAI the option to acquire up to 10% of AMD’s stock.
The news sent AMD shares soaring over 30%, underlining OpenAI’s growing ambitions to diversify beyond Nvidia and secure vast computing power for the next era of AI innovation.
