An internal strategy document from OpenAI has surfaced, which details plans to transform the company’s ChatGPT into a full-blown “Super-Intelligent Assistant” as early as the first half of 2025.
The document, said to have been revealed during the ongoing antitrust trial of US Department of Justice against Google, is about OpenAI’s plans to evolve ChatGPT from a generalist chatbot into an AI companion that can have deep knowledge about you and can truly be a part of your life in a very holistic manner.
The Super Assistant would be an “intelligent entity with T-shaped skills,” OpenAI writes in the leaked document, which is called, “ChatGPT: H1 2025 Strategy.” Its base would be extensive functionality for everyday use, but the brain-in-a-box would also have deep capabilities in harder-to-master domains.
According to the document, this souped-up version of ChatGPT would be able to perform all manner of tasks, from answering simple questions and organizing calendars to more complicated endeavors such as helping people find a place to live, hiring a lawyer, arranging vacations and providing emotionally intelligent assistance.
It will become increasingly apparent as we develop new models, o2 and o3, that the technologies underlying these models are sufficiently sophisticated to carry out agentic tasks in a robust manner.
Second, inclusion of tools such as computer use, a long with progress in multimodal interaction and generative user interfaces, will greatly improve ChatGPT’s capability to perform actions and interact with users in more natural ways.
The company’s vision goes beyond today’s shapes, such as websites, phone apps and desktop software. The document suggests a future in which ChatGPT is increasingly omnipresent, helping users at home by running errands, answering questions, playing music and suggesting recipes, while also helping them on the go through navigation, restaurant recommendations and calling up friends.
As a companion conversation tool In a work setting, this might be taking meeting notes or pitching in on that slideshow draft, or just offering companionship and processing as people go about solo activities.
Remarkably, the document also talks about OpenAI’s hardware strategy. That vision of a more cohesive AI assistant also fits with not one, but two major collaborations OpenAI CEO Sam Altman recently hinted at—one with former Apple design chief Jony Ive to build new AI hardware.
The exact nature of this hardware is still unknown, but there’s speculation it might be some kind of smart home product that would integrate the Super Assistant even more tightly into your life.
The leaked document does also concede in its own way there will be challenges, most notably against the might of established tech behemoths who could, through their existing reach to consumers, push their own AI assistant.
Here OpenAI plans to lobby for laws regulating peoples’ right to choose ChatGPT as their default assistant on different devices. Space and infrastructure constraints from ChatGPT’s growing user base are also understood, and the company is aggressively upgrading its data center capacity with projects such as “Stargate.”
The fact that such grand ideas have emerged from the secretive startup shows OpenAI’s determination to continue to expand the frontier of artificial intelligence and to revolutionize the ways that people interact with machines. We can only wait and see in the next few months if this vision really comes true and how it would change the competitive play of AI assistants.