Tech • AI • Robotics • Game
Build What’s Next on the ChatGPT Platform with Plugins | DevDay 2026
9.3/10TL;DR
ChatGPT is expanding its plugin platform with new discovery tools, native in-app extensions, easier deployment, and automated reviews to help developers build and distribute more capable integrations.
KEY POINTS
Discovery moves into the chat
ChatGPT can now suggest relevant plugins directly inside a conversation when a user’s task matches a tool’s capabilities. Examples included recommendations for Supabase and Vercel when a user asks how to turn an existing website into a full-stack web app, creating a new path for plugin installation during real work rather than through browsing alone.
Ranking is tied to usage quality
Plugin visibility is shaped by three core signals: conversion, active usage, and retention. The system favors plugins that users install, use regularly on a daily or weekly basis, and return to repeatedly, with the goal of surfacing products already proving useful in practice.
The directory has grown to 4,000 plugins
The plugin ecosystem now includes 4,000 plugins in the directory, reflecting broader adoption across development, design, research, and data-related work. Installations have climbed sharply over the past six months as support and reliability improved and as developers aligned around a shared agent plugin format.
Plugins are becoming central to agent workflows
More conversations now involve plugins as users assign agents more complex, multi-step tasks. Those tasks increasingly require access to external context and the ability to take actions across multiple systems, driving strong growth in MCP, which is being positioned as a core way agents interact with software and data.
New plugin extensions add native interfaces
A major update introduces plugin extensions, allowing developers to bring their products more directly into ChatGPT across web, mobile, and desktop. Plugins can now appear in the sidebar as a persistent home or in a side panel beside the conversation, letting users inspect results and make changes without leaving the chat.
Shared context powers collaborative workflows
The new interface model allows a plugin and the model to share what a user is viewing or selecting in real time. Developers can also add targeted actions in the composer, support at-mentions for specific tools or content, and register plugins to open files in interactive viewers inside ChatGPT.
Internal and partner examples show broader use cases
OpenAI teams have already used the extension platform for health, finance, security, code review, and meetings. A meetings plugin can transcribe notes, summarize discussions, pull in context from other systems, and launch follow-up action threads. In life sciences, Rosalind Workbench lets researchers inspect molecular structures in a 3D viewer while the conversation stays aware of the selected region.
Third-party apps are building native experiences
Design tool Magic Path demonstrated voice-driven page creation directly inside ChatGPT, generating a landing page while rendering the result in a live sidebar canvas. Other examples included Atlassian for Jira and Confluence, Canva for pitch decks, and Shopify for store analytics, product bundles, image generation, and checkout workflows.
Sign in with ChatGPT is aimed at reducing friction
A new Sign in with ChatGPT option lets developers add identity integration to their web apps and plugins. Supabase reported three times the weekly plugin installations after adding the feature, and eligible users can also bring certain ChatGPT token entitlements into partner products, potentially reducing the need for separate AI charges.
Building and shipping are being simplified
Developers can build with Codex, host on Sites, and use tools such as App Plugin Creator to package and deploy hosted MCP servers. Submissions are also being overhauled: the old form is being removed, required metadata now lives in plugin.json, review checks are increasingly automated at upload, and MCP server updates can go live without resubmitting a new plugin version.
ChatGPT adds support for MCP events
ChatGPT is also adding support for the upcoming MCP event spec, enabling servers to notify the platform when something changes and trigger automations even when a user is away. The flow relies on event listing, subscription, and signed webhooks, creating a simpler route for developers to build event-driven automations.
CONCLUSION
The latest plugin changes push ChatGPT beyond a chat interface toward a more integrated work platform. The main test now is whether easier discovery, stronger native interfaces, and a lighter submission process translate into sustained developer adoption and higher-quality tools.
Ask a question