With the release of ChatGPT one year ago, OpenAI introduced the world to the idea of an AI chatbot that seems to do anything. Now, the company is releasing a platform to create custom versions of ChatGPT for specific use cases – no code required.
In the coming weeks, these AI agents, which OpenAI calls GPTs, will be available through the GPT Store. Details about how the store will look and function are not yet available, although OpenAI promises that it will eventually pay creators an undisclosed amount based on how much of their GPTs are used. GPTs will be available to paying ChatGPT Plus subscribers and OpenAI business customers, who can only create internal GPTs for their employees.
The custom GPTs were announced Monday at DevDay, OpenAI’s first-ever developer conference in San Francisco, where the company also announced a turbocharged, low-cost, low-cost GPT-4 for developers to use its models in their applications, along with news that ChatGPT has reached 100 million weekly users.
“Since the introduction of ChatGPT, people have been asking for ways to customize ChatGPT to fit the specific ways they use it,” said OpenAI in a statement shared with The Verge. “We introduced Custom Instructions in July that allow you to set specific preferences, but requests for more control have kept coming. Many power users keep lists of carefully crafted commands and command sets, and copy them into ChatGPT. Now GPTs do all that for you.”
During a recent demo I received of OpenAI’s GPT platform, a bot called “Creative Writing Coach” critiqued an uploaded PDF of a writing sample. In about two minutes, I watched another GPT being turned around to help attendees navigate DevDay. The platform automatically named the bot “Event Navigator,” generated a profile picture for it using DALL-E, and included a PDF attachment with an event schedule to inform its responses.
The OpenAI interface allows you to direct how you want GPT to interact with people before you publish. The DevDay Event Navigator agent I saw during my demo was instructed to be helpful and concise and avoid scheduling conflicts. OpenAI has automatically generated initial chat commands, such as “What’s the first session today?”
Each GPT can be given access to web browsing, DALL-E, and OpenAI’s Code Interpreter tool for writing and running software. There is also an “Info” section in the builder interface for uploading custom data, such as the DevDay event schedule. With a feature called Actions, OpenAI allows GPTs to tap into external services to access data such as emails, databases, and more, starting with Canva and Zapier.
The introduction of custom GPTs means that OpenAI is now competing with other AI bot platforms such as Character.AI and Meta, which recently launched its AI bots on WhatsApp, Instagram, and Metassenger. OpenAI positions its platform as more focused than its competitors, rather than emphasizing bots that act like humans, although it doesn’t argue against people building GPTs with humans.
OpenAI is now competing with other AI bot platforms such as Character.AI and Meta
The creators of GPTs won’t be able to view the conversations people have, and it’s unclear what high-level usage data they’ll have access to. OpenAI says it will be monitoring work to prevent things like fraud, hate speech, and “adult themes.” When the GPT Store launches down the road, OpenAI will only accept agents from people who have verified their identity. Initially, GPTs will be accessible through shared web links.
Ultimately, OpenAI sees its GPT platform as bringing it one step closer to its ultimate goal: the creation of AI superintelligence, or AGI. Limiting access to paid subscribers should also help boost profits for the already accelerating company as it reportedly seeks to raise as much as $90 billion from investors.