OpenAI intends to create a specialized agent, Agentic Software Engineer (A-SWE), which, using artificial intelligence, can perform software development tasks and test it for errors.
OpenAI intends to create a specialized agent, Agentic Software Engineer (A-SWE), which, using artificial intelligence, can perform software development tasks and test it for errors.
«The third (product) coming is something we call A-SWE. By the way, we’re not the best marketers, you might have noticed, but [it’s] a software developer called Agentic,» Officechai writes, citing OpenAI CFO Sarah Friar during her conversation with Goldman Sachs.
She added that A-SWE will not be a typical AI assistant like current LLMs like Copilot. «It’s literally a software development agent that can build an app for you. It can take a PR [pull request] that you would give to any other engineer and go build it,» Sarah Friar noted, emphasizing that the new AI agent will be a fully autonomous developer.
In addition to developing software itself, OpenAI’s new AI agent will be able, according to the CFO, to perform other tasks that developers «hate to do,» including:
«Something you can never force on software engineers,» Sarah Friar added regarding the last point.
OpenAI CEO Sam Altman previously stated that by the end of 2025, the world’s best programmer will be an artificial intelligence.
Recall, in early April, OpenAI announced the attraction of $40 billion in investments to stimulate artificial intelligence research.
