AI agents are remodel how software is written, scale and experienceand Many expect The technology for Unlock profits IA companies have promised for a long time. While most companies today stay in the “test” phaseAs the agents make their way throughout the organization, workers must discover how integrate them into their workflows. That is particularly true for developers, who can use agents to increase efficiency and, in many cases, will also be responsible for building, maintaining them and integrating them.
Agents are autonomous programs that depend on underlying models such as language models or planning systems that are capable of executing tasks without constant human orchestration. (As Chip Fleen has pointed out, many consider them “The final objective of AI. “) It may sound obvious, but what this distinguishes as a novel approach is the” agency “: to operate independently according to the objectives, memory and pre -established tools.
Agents can be simple, make a single call API based on the user or complex entrance, which orchestrate multiple services, collaborating with other agents and learn over time. But they will only be as useful as the data and systems to which they connect, and that means that APIs will continue to play a huge role. As the bridge between the agents and the digital world, the API make it possible for AI agents to access data, perform actions and integrate with external systems to achieve their objectives. But what does it mean to build for a world where agents, fed by API, act on their own?
APIs are not a new technology; The concept dates back to the 1940s. And the AI has not changed the objective of a well thought out API: Easily deliver valuable functionality to third parties. However, traditional APIs have always been designed with human developers in mind. API compatible with the agent do not have the same requirements. For APIs to effectively serve agents, they must be consumed from the machine, self -writing and semantically rich. This requires that developers prioritize clear functionality, descriptive metadata and the management of real -time errors, all while maintaining accessibility for human users. There are also new protocols to consider, including the model context protocol (MCP) and the Agent2agent (A2A) protocol, which can be used to communicate with external data sources, tools and other agents.
The API do not go soon, but developers try to optimize their systems and software must learn the new protocols that will help them connect agents with their systems and data. They should also consider the technical environment in which their APIs now circulate and design for both humans. and Agents There is no time like the present to begin.
Do you want to learn more? Join the host Mike Amundsen and an estimated line of API experts on July 17 for the O’Reilly Api Superstream, all about the creation of API optimized for AI agents. More than four hours full, will explore problems with current APIs; How to integrate your API with AI and MCP; Business agent ecosystems; The synergy between API, LLMS and XAI; Azure API management; And much more. It is free for O’Reilly members. Register here.
Isn’t it a member? Register for a free 10 -day test To attend, and see all the other great resources in O’Reilly.
#developers #OReilly
