OpenAI facilitates new tools to create apps from agents who sail on the web and use the computer

Bringing agents the most recent capabilities of OpenAi models requires “thorough iteration and a personalized orchestration logic without sufficient visibility or integrated support”, which makes work difficult.

The technological firm has responded to this recurring complaint among its customers with a set of application programming interfaces (API) and tools “specifically designed to simplify the development of agent applications.”

In the center of these novelties is the response API (RESPOS API), which offers a basis for using openai models and tools integrated in the creation of agents. In this way, they can access web navigation and file search, and use the computer autonomously to perform tasks.

This is accompanied by a new development kit (SDK) of agents, which helps to orchestrate workflows both in multiple agents, and tools that allow following its execution.

“As the model’s capabilities become increasingly specific, we will continue to invest in deeper integrations in our API and new tools to help implement, evaluate and optimize the agents in production,” said the technological firm in its official blog.