Amp is building a frontier AI agent that helps developers turn ambitious ideas into working software. According to CEO Quinn Slack, each improvement in OpenAI's models is an opportunity to expand what the agent can do, from helping developers decide what to build to writing code and designing pages. Amp selects models based on what works best for its users, and OpenAI models account for the majority of tokens used in Amp. Slack praised OpenAI's transparency about safeguards, saying users understand them and can do all their work in Amp with OpenAI models.
As builders give agents more work, token costs can become a constraint. Slack describes some Amp users consuming 10 billion tokens a week, and he believes watching every increment of usage could keep people from discovering what an agent is capable of. To address this, Amp launched its Sign in with ChatGPT integration, which gives users access to more tokens and OpenAI models, including GPT-6 Astra. Slack says this lets people stop counting every token and let agents take on longer tasks, plan, decide what to build, and keep working after they ship.
Amp also uses agents across its own operations. Its 20-person team relies on an agent for marketing, improving the experience for new users and helping more people find Amp through search and other channels. GPT-6 Astra helps produce documentation and design pages that meet the company's standards. Slack also experiments at home, using the model to build a 3D house model in Blender that he and his children turn into games.
Amp recently announced Puck, a conversational assistant that launches and manages agents. Through Calling Puck, powered by GPT-Live-1, users can speak to it in real time, even from a watch. Amp is also testing Ultrafast mode with GPT-6 Astra. Slack says the greater speed could reduce how many agents people juggle at once, or give agents more time to check their own work before returning results.
Source: OpenAI Blog · Summarized by HeadlinesBriefing