OpenAI releases API tools for developing AI agents – A milestone in the development of autonomous AI systems – Image: Xpert.Digital
OpenAI sets new standards in the development of autonomous AI systems
Autonomous AI made easy: OpenAI launches innovative APIs and tools
Yesterday, March 11, 2025, OpenAI unveiled a comprehensive suite of new tools and APIs specifically designed for developing autonomous AI agents. This release marks a significant step forward in democratizing agent development and represents a strategic move to empower developers to create powerful, self-acting AI systems. The new tools provide an enhanced foundation for developing agent-based applications and could fundamentally transform how businesses and developers interact with AI technology.
Related to this:
The Responses API: The heart of the new agent platform
At the heart of yesterday's announcement is the new Responses API, which provides a more flexible foundation for developing agent-based applications. This API cleverly combines the simplicity of the Chat Completions API with the advanced tool capabilities of the Assistants API. For developers, this means a significant simplification of the development process, as they can now build more complex applications with less effort. The Responses API is designed as a long-term replacement for the existing Assistants API, with OpenAI planning to discontinue the latter by mid-2026 after the complete integration of all features into the new API.
The strategic realignment towards the Responses API demonstrates OpenAI's commitment to streamlining the development process for agent-based applications. While the existing Chat Completions API will remain available for developers who don't require integrated tools, OpenAI recommends migrating to the new Responses API, as it offers the same performance with added functionality. This API consolidation reflects the desire to create a more cohesive development environment while simultaneously expanding opportunities for innovative applications.
Integrated tools for autonomous agents
A standout feature of the new platform is its three integrated main tools, which provide agents with enhanced capabilities. The web research tool is based on the same models that ChatGPT uses for its search functions: GPT-4o-search and GPT-4o-mini-search. These models achieve impressive accuracy scores of 90% and 88%, respectively, for factual queries in the SimpleQA benchmark and provide answers with source citations, similar to ChatGPT. Pricing for this tool is $30 per thousand queries for GPT-4o-search and $25 per thousand queries for the mini version.
The second integrated tool, file search, allows developers to extract relevant information from extensive document collections. This tool handles parsing, chunking, and embedding the data, providing a ready-to-use vector storage solution that can be used for $2.50 per thousand searches plus storage costs of $0.10 per gigabyte per day, with the first gigabyte being free.
As its third and perhaps most innovative tool, OpenAI introduces the Computer-Using Agent (CUA), based on the same model as the already familiar AI agent "Operator." This CUA allows AI agents to interact with computers by interpreting graphical user interfaces and interacting with them like a human—by clicking, scrolling, and typing. Despite its impressive capabilities, the CUA currently achieves a success rate of 38.1% on the OSWorld benchmark, 58.1% on WebArena, and 87% on WebVoyager for web-based interactions when performing complex computing tasks. Using this tool costs $3 per million input tokens and $12 per million output tokens, but is currently only available to select developers in higher usage categories.
The Agents SDK: Orchestration of Multi-Agent Systems
Another key component of yesterday's announcement is the new open-source Agents SDK, which enables the orchestration of both single and multiple agents. This SDK represents an evolution of the earlier experimental Swarm framework, which, due to its popularity, has now been upgraded to a fully-fledged, officially supported development tool. It offers four core features: easily configurable LLMs with clear instructions and integrated tools, intelligent agent handoffs, configurable security controls for input and output validation, and tracing and observability tools for visualizing and optimizing agent execution.
The Agents SDK explicitly supports a number of common agentic patterns, such as workflows, handoffs, agents-as-tools, LLM-as-a-judge, parallelization, and guardrails. It is currently available for Python, with support for Node.js coming soon. A particular advantage of the SDK is its compatibility not only with OpenAI's own APIs but also with third-party models, provided they offer an API that conforms to the Chat Completions standard. This openness facilitates integration into various development environments and supports flexibility in choosing the underlying models.
Practical applications and initial success stories
The first practical use cases for OpenAI's new agent-based tools are already emerging. Companies like Hebbia and Coinbase have successfully implemented the new tools, demonstrating early confidence in the capabilities of these technologies. This focus on practical applications shows a deliberate shift away from mere concept presentations towards genuinely usable solutions that can offer real added value.
OpenAI is also already collaborating with internet companies such as DoorDash, Instacart, OpenTable, StubHub, and Uber to develop customized agent solutions for their platforms. These partnerships underscore the potential of agent-based AI to optimize processes across industries and create new user experiences. The versatility of applications ranges from placing food orders and making restaurant reservations to performing complex research and analysis tasks.
The future landscape of agentic AI
With yesterday's release, OpenAI strategically positions itself in the growing competition for dominance in the field of agentic AI. Competitors like Google and Anthropic are also working on similar technologies, with projects such as Google's DeepMind Mariner and Anthropic's Computer Use pursuing comparable approaches. What distinguishes OpenAI's solution, however, is its ability to interact directly with graphical user interfaces without requiring specific integrations.
The development of agent-based AI aligns with broader trends in the AI industry. A growing number of companies are relying on so-called "reasoning models" that can solve complex tasks and record their thought processes. These models form the basis for autonomous agents that can be deployed across various industries. OpenAI is also apparently planning to develop even more specialized agents tailored to specific fields such as software development or scientific research.
Related to this:
Impact on the developer community and companies
OpenAI's new tools address key challenges developers face when building AI agents. These include optimizing prompts, coordinating complex workflows, and, most importantly, bringing transparency to the development process. By providing integrated monitoring tools, OpenAI makes agent workflows traceable, which is crucial for both debugging and optimization.
Olivier Godement, OpenAI's product manager, emphasizes the challenges of scaling agents: “It's fairly easy to demonstrate your agent. However, scaling an agent is quite difficult, and it's very difficult to get people to use it frequently.” This statement underscores the need for new tools designed to simplify the development process and encourage the creation of reliable, user-centric agents.
Future prospects and next steps
OpenAI plans to add more tools and features in the coming weeks and months to simplify and accelerate the development of agent-based applications on its platform. The company is convinced that agents will soon become an integral part of the workforce and significantly boost productivity across various industries. “We believe that agents will soon be an integral part of the workforce and will significantly increase productivity across industries,” OpenAI stated in a blog post.
Yesterday's release of the agentic tools is just the beginning of a longer-term strategy. Meta's Head of Business AI, Clara Shih, even predicts: "We are rapidly approaching a point where every company, from very large to very small, will have a business agent representing it and acting on its behalf, in its voice—just as companies have websites and email addresses today." This vision demonstrates how transformative agentic AI could be for the business world.
A milestone in the development of autonomous AI systems
Yesterday's release of agentic tools by OpenAI marks a significant milestone in the development of autonomous AI systems. By combining powerful APIs, integrated tools, and a flexible SDK, OpenAI has significantly lowered the barrier to entry for developing AI agents. This could lead to accelerated adoption of this technology across various industries and fundamentally change the way we interact with AI.
The strategic significance of this release lies not only in the technical capabilities it offers, but also in the vision it represents: a future where autonomous AI agents act as reliable assistants in professional and everyday life. While the technology is still in its infancy, yesterday's announcement clearly demonstrates OpenAI's determination to play a leading role in shaping this future. The coming months will reveal how developers and businesses will leverage these new tools to create innovative solutions that unlock the full potential of agent-based AI.
Related to this:
Your global marketing and business development partner
☑️ Our business language is English or German
☑️ NEW: Correspondence in your native language!
Konrad Wolfenstein
I and my team are happy to be available to you as your personal advisor.
You can contact me by filling out the contact form here wolfenstein@xpert.digital:or simply call me at +49 7348 4088 965. My email address is
I'm looking forward to our joint project.
