OpenAI unveils GPTs, letting anyone build custom ChatGPTs with zero code

Until now, making ChatGPT behave like a dedicated specialist meant hoarding a notepad full of intricate prompts and pasting them into every fresh conversation.

OpenAI just changed that with GPTs, rolling out custom versions of ChatGPT that anyone can build tailored for specific tasks, work routines, or everyday hobbies.

Building an app just by chatting

Creating a GPT requires zero coding. You literally open up a conversation, explain what you want the assistant to do, and the system sets it up for you.

You can feed it custom instructions, upload reference documents like PDFs, and toggle built-in skills including web browsing, DALL-E image generation, and data analysis. If you need it to hook into external services, you can define Actions using external APIs, turning the assistant into something that can query private databases, interact with your inbox, or trigger e-commerce workflows.

It is available right away for ChatGPT Plus and Enterprise subscribers, with sample builds from Canva and Zapier already live. Enterprise workspaces get access on Wednesday, where companies like Bain, Amgen, and Square are already deploying internal-only bots for brand copywriting and engineering onboarding.

OpenAI is also launching the GPT Store later this month, featuring verified builders, category leaderboards, and a revenue-sharing program based on usage coming down the pipeline.

Alongside the announcement, ChatGPT Plus dropped one of its most annoying quirks: manual model switching. You no longer have to toggle between browsing, DALL-E, and Advanced Data Analysis. The assistant now routes tools automatically, accepts document uploads directly, and extends its knowledge base to April 2023.

Why it matters

OpenAI is pulling off its own App Store moment, shifting from a pure model vendor into an end-to-end platform.

For everyday creators and domain experts, the barrier to building functional software just vanished. If you possess deep niche knowledge, you no longer need an engineering team to turn your playbook into a usable tool. That significantly shifts power toward subject-matter experts.

For the AI startup ecosystem, however, this launch is a wrecking ball. Over the past year, hundreds of startups raised capital just to wrap system prompts and basic document retrieval into clean user interfaces. GPTs make those single-feature products redundant overnight.

That said, we think building a true agent ecosystem will not be effortless. Conversational interfaces have obvious usability limits for complex workflows, and without robust quality control, the GPT Store could quickly turn into an ocean of low-effort prompt spam. OpenAI now has to prove it can run a healthy software marketplace, not just train frontier models.