OpenAI has launched a suite of developer tools designed to help companies create AI agents that can perform tasks on behalf of users.
The announcement this week aims to making agent development easier, while addressing common challenges in turning advanced capabilities into working applications.
The toolkit arrives amid broader developments for the company, including its recent Sora video generation system launch and reports of premium pricing tiers, as OpenAI continues to expand its market presence across multiple AI domains.
Tools for Agents Across All Industries
The announcement features four main components:
- Responses API: This combines elements of the Chat Completions API with the tool capabilities of the Assistants API. Developers can now solve complex tasks using multiple tools with a single API call.
- Built-in Tool Options: The platform now offers web search, file search, and computer use capabilities. These tools work together to connect AI models with real-world data and systems.
- Agents SDK: This open-source toolkit helps manage multi-agent workflows. It includes features for configuring AI models with clear instructions, transferring control between agents, implementing safety checks, and visualizing how agents work to aid debugging.
- Observability Features: New tracking tools help developers understand what happens during agent execution, making it easier to identify and fix problems.
In their annoucement, OpenAI noted:
As model capabilities become more and more agentic, we’ll continue investing in deeper integrations across our APIs and new tools to help deploy, evaluate and optimize agents in production. Our goal is to give developers a seamless platform experience for building agents that can help with a variety of tasks across any industry.
Computer Control Capabilities
The Computer-Using Agent (CUA) model that powers the computer use tool has shown promising results in testing. The model achieved a 38.1% success rate on OSWorld for general computer tasks, 58.1% on WebArena, and 87% on WebVoyager for web-based interactions.
This tool captures mouse and keyboard actions from the AI model; allowing developers to automate tasks by converting these actions into commands that work in their systems. Companies like Unify have used this capability to access information that wasn’t previously available through programming interfaces, while Luminai has applied it to automate workflow processes for organizations with older computer systems.
A World of Possibilities: Business Applications
The ability to design and deploy AI agents could offer advantages across various sectors. Based on the capabilities described in OpenAI’s announcement, these tools might enable applications such as:
- Healthcare: Medical facilities could potentially create agents to handle appointment scheduling and preliminary patient information gathering, which might free up staff for direct patient care.
- Finance: Banks and investment firms might deploy agents to monitor market conditions and handle routine client inquiries about account status.
- Manufacturing: Production facilities could potentially use agents to monitor equipment status and predict maintenance needs based on operational data.
- Retail: Businesses might implement agents that provide personalized shopping assistance and handle inventory management across multiple channels.
- Legal: Law firms could utilize agents to conduct preliminary document reviews and organize case materials.
- Education: Educational institutions might create agents that assist with administrative tasks and help monitor student progress across learning platforms.
The economic impact could also extend beyond efficiency.
Companies implementing these agent systems might potentially reduce error rates in routine tasks and redirect human talent toward creative and strategic work that requires human judgment.
Early Adoption Examples
Several companies have already put these tools to use:
- Coinbase have built AgentKit to help AI agents interact with cryptocurrency wallets and blockchain activities.
- Box have created agents that search through both internal Box documents and public internet sources to provide insights while respecting security permissions.
- Hebbia helps financial firms and law practices extract insights from large datasets using the web search capabilities.
- Navan uses file search in its travel assistant to provide answers based on company-specific policies.
OpenAI’s Ups and Downs
While OpenAI focuses on agent development tools, the company has been active in several other areas. Reports suggest potential pricing for professional agent services ranging from $2,000 per month for knowledge workers to $20,000 per month for research-focused applications.
These pricesdrew criticism and could create openings for competitors like Deepseek to offer less expensive alternatives.
On the partnership front, Microsoft has reportedly been testing its own AI models (called MAI) that perform similarly to OpenAI’s offerings. Although the two companies maintain their partnership through 2030, Microsoft appears to be keeping its options open by developing alternative AI systems.
OpenAI has also expanded into video generation with Sora, launched shortly after Alibaba’s Wan 2.1 system. Sora allows users to create videos from text descriptions, with tools for editing, extending, organizing, and combining video clips.
The system integrates with OpenAI’s existing subscription plans, with ChatGPT Plus ($20/month) offering basic video generation and ChatGPT Pro ($200/month) providing higher resolution and additional features.
Building the Agent Economy: Tools for Everyone
OpenAI’s agent tools are a step toward making AI more useful for practical tasks. As companies continue exploring how AI can handle complex workflows, these development tools could help bridge the gap between experimental AI capabilities and practical business applications.
For an industry often criticized for creating tools that replace human labor, OpenAI’s agent development platform may ultimately prove most valuable when it enables people to build systems that amplify human capabilities rather than merely automate existing workflows
