~4m18:56OpenAI DevDay: Dots, Agents & $100B Opportunities
Sep 29, 2026
Read: ~4m · You save: 15 min
OpenAI DevDay: Dots, Agents, and $100 Billion Opportunities
OpenAI DevDay: Sam Altman unveils Dots, Agents API & ChatGPT login. Discover how to earn billions from new AI opportunities!
The OpenAI DevDay 2026 event introduced a series of innovations that, according to experts, are capable of opening new horizons for business and leading to the emergence of multi-billion dollar markets. Among the more than 20 announced features, particular attention was drawn to platforms for personal agents, an API for decision-making, and the integration of ChatGPT login.
The "Dots" Personal Agent Platform and Plugin Ecosystem
The key announcement was the "Dots" personal agent platform, positioned as OpenAI's response to similar developments from Meta (Muse) and other companies. "Dots" provides users with access to ChatGPT's 1.2 billion weekly active users. Users will be able to set tasks, such as "fix a building permit" or "find a restaurant with a specific cuisine," and the system will select appropriate plugins to complete them.
OpenAI has reaffirmed its commitment to developing plugins, despite previously mixed feedback. The company aims to transform ChatGPT into a comprehensive artificial intelligence application ecosystem. For businesses, this means the opportunity to create plugins that will solve specific customer problems, whether it's submitting permits or providing search results. Successful plugins will gain access to a vast user base, opening up the potential for billions in transaction volume within the platform. In the long term, the emergence of trillion-dollar markets related to personal agent platforms is predicted.
Decisions API: Intelligent Real-Time Decision Making
The presented Decisions API is OpenAI's response to Jev-like solutions. This API allows developers to leverage the intelligence of the Luna model for real-time decision-making based on given questions with a limited set of predefined answers. Developers can provide context in the form of text or images and receive responses that can be used for content classification, request routing, or determining an agent's next action. The API is currently in limited preview.
One potential application of the Decisions API is user interface optimization, such as creating the most conversion-driving screenshots for app stores or landing pages. The question of whether Decisions API will surpass Jev in its capabilities remains open and will be determined as it further develops and is utilized.
Agents API: Empowering Software Interaction
Agents API now supports interaction with computer software, enabling developers to build agents capable of performing tasks by interacting with applications. This feature includes Codex's capabilities for multi-agent collaboration, tool discovery, tool invocation, and context compression. OpenAI manages the infrastructure, allowing development teams to focus on building applications.
This opens up possibilities for creating vertical solutions, where companies with unique data, strong brands, or deep industry expertise can develop applications powered by OpenAI's managed infrastructure. Agents API is seen as a new surface for building "AI-native" applications.
Login via ChatGPT: A New Era for Business Models
A significant, though less publicized, announcement is the ability to log into third-party websites using ChatGPT credentials and existing token limits. This feature, similar to the launch of "Login with Facebook," has the potential to generate thousands of new business opportunities and lead to the creation of multi-billion dollar companies.
The new model allows for the creation of free applications, expanding the user acquisition funnel. Integration with ChatGPT enables the use of existing user subscriptions, lowering the barrier to entry. The core functionality of an application can be free, with monetization occurring through the sale of additional features, enterprise solutions, enhanced collaboration capabilities, or access to more data. This eliminates the need for inference payments for basic functions, allowing startups to focus on distribution, branding, and data collection.
Examples of potential applications include:
- An agent for cleaning architectural CAD files.
- A research tool for monitoring a specific scientific field.
- An application for optimizing Shopify catalogs.
- A video production tool with a repeatable editing workflow.
- A contract verification system for a specific type of franchise agreement.
This model makes niche categories economically viable that were previously too small for OpenAI's standalone AI features. The focus shifts from selling tokens to selling complete workflows. Key elements for building a business in this new paradigm include: proprietary information, specialized workflows, access to external systems, collaboration and auditing capabilities, as well as networks of providers and measurable business outcomes.
The "Trigger-Decision-Action-Feedback" Model
To build a business in the new AI environment, a four-stage model is proposed:
- Trigger: An event that initiates an action (e.g., invoice payment deadline, low inventory level, tax period).
- Decision: Determination of further steps by the ChatGPT system.
- Action: Task execution by an application or agent.
- Feedback: The result of execution (approved, rejected).
The most stable businesses will control two out of three elements: trigger, action, and feedback.
New Business Ideas
Real-World API
As the number of users turning to "Dots" with various requests grows, the need for integration with the real world arises. Creating a platform that connects verified specialists (e.g., permit expediters, customs brokers) with user requests can be a significant opportunity. Such platforms can earn commissions on transactions, bridging the "last mile" in the real world.
Analytics and Optimization for Agent Discovery
With the emergence of numerous personal agent platforms (Dots, Muse, Gemini, Instinct, etc.), there is a growing need for tools that help developers ensure the visibility of their plugins. An analytics platform capable of generating realistic prompts, measuring plugin activation rates based on customer intent, identifying issues with metadata and tool definitions, and tracking task completion success could become highly sought after. Analogous to the web era, which saw the rise of Mixpanel, Amplitude, and Google Analytics, the era of personal agents is expected to see the emergence of similar products for optimizing the discovery and utilization of AI agents.
Other Announcements
Among other important, but expected, announcements:
- GPT 6.1 Soul: A significant update to the GPT-6 model.
- Ultrafast: An accelerated version of the model, providing faster performance.
- Private Intelligence: Solutions for protecting company data when using advanced AI models.
- Spaces: A tool for collaborative work with AI agents in cloud documents and spreadsheets.
These innovations, especially the Decisions API, Agents API, ChatGPT login, and the "Dots" platform, represent significant steps in AI development, opening up new opportunities for entrepreneurs and developers.
Introduction: OpenAI's Revolutionary Launches
Sam Altman announced 20+ new features, among which three or four stand out as capable of generating billions of dollars. The main focus is on building businesses around the new capabilities of ChatGPT.
- OpenAI DevDay introduced over 20 new features.
- Several key launches have the potential to create businesses with billions of dollars in revenue.
- The primary goal is to help people leverage the new features for profit within the ChatGPT ecosystem.
Dots: Personal Agent Platform
Introducing Dots, a personal agent platform that competes with Muse and Instinct. It integrates with ChatGPT's 1.2 billion active users, allowing plugins to perform user tasks and transform ChatGPT into an app ecosystem.
- Dots is a personal agent platform from OpenAI.
- It is a response to Meta's Muse and Instinct.
- Dots provides access to ChatGPT's 1.2 billion weekly active users.
- Plugins (mini-apps) will perform user tasks, such as finding restaurants or assisting with documents.
- OpenAI is betting on the development of plugins and turning ChatGPT into an app ecosystem.
Decisions API: Real-time Decision-Making
Introducing the Decisions API—an alternative to Jev designed for real-time decision-making. It empowers developers to leverage AI for content classification, request routing, and agent action selection.
- Decisions API is OpenAI's answer to Jev.
- The API is designed for real-time decision-making.
- Developers can use it to classify content, route requests, and determine agent actions.
- The API is in limited preview.
Agents API: Computer Vision Integration
The Agents API now supports computer vision, enabling the creation of agents that interact with software to perform tasks. This includes Codex capabilities: tool discovery, tool invocation, and context compression.
- The Agents API now supports computer vision.
- Developers can create agents that interact with software.
- The API includes Codex capabilities: multi-agent capabilities, tool discovery and invocation, and context compression.
- OpenAI manages the infrastructure, allowing teams to focus on application development.
Login via ChatGPT: New Business Models
The ability to log in via a ChatGPT account has emerged, opening up thousands of new business opportunities, similar to the launch of Facebook login. This allows for the creation of free basic applications that can be monetized through additional features.
- It is now possible to log in to websites using a ChatGPT login.
- This opens up thousands of new business opportunities, similar to the previous Facebook login.
- Allows for the creation of free basic applications, attracting more users.
- Monetization is possible through the sale of additional features, corporate solutions, or data.
- Eliminates the need to pay for inference for basic application features.
AI Era Business Building Model
A new business-building model for the AI era is presented: trigger, decision, action, and feedback. The most successful businesses will own two of the three components: trigger, action, and feedback.
- A 4-stage model for business building is proposed: trigger, decision, action, feedback.
- Trigger: An event that initiates an action (e.g., invoice due date).
- Decision: AI determines what to do next.
- Action: An application performs a task.
- Feedback: The result of the action (approved/rejected).
- The strongest businesses will control the trigger, action, and feedback.
Promising Business Ideas: Real World and Analytics
Two key business ideas are considered: creating real-world APIs that connect AI with physical services (e.g., permit processing), and developing analytical platforms for discovering and optimizing agents.
- Idea 1: Real-world APIs connecting AI with physical services (e.g., permit expediters, customs brokers).
- Idea 2: Analytical platforms for discovering and optimizing agents (similar to Google Analytics in the web era).
- These platforms will track plugin performance, analyze user intent, and identify issues.
- There is a huge opportunity to create niche products that are not core OpenAI features.
Other Important Launches and Takeaways
Other launches are briefly mentioned: GPT 6.1 Soul, ultrafast (an accelerated version of the model), and Private Intelligence (for data protection). The main focus is on Dots, Decisions API, Agents API, and login via ChatGPT as key drivers for the future.
- GPT 6.1 Soul launched — a major update to GPT-6.
- Ultrafast introduced — a significantly faster version of the model.
- Private Intelligence aims to protect company data when using AI.
- Spaces — an analog of cloud documents for collaborative work with AI agents.
- Dots, Decisions API, Agents API, and login via ChatGPT are key for business.