Meta Introduces Muse Code: An AI Tool for Efficient Management of Large Codebases

Meta, often perceived as a newcomer in the AI tools landscape, is making significant progress to catch up. The company has recently introduced a new coding assistant designed for developers needing help with intricate tasks in extensive software codebases.

Currently in beta, Muse Code is capable of executing “complete software engineering tasks across large repositories,” as highlighted by Meta CEO Mark Zuckerberg in a social media update on Wednesday. These functions include “planning modifications, writing code, and validating results,” he mentioned.

This coding assistant can be easily set up with a single command and utilizes Meta’s previous coding model, Muse Spark. It handles large projects by initiating its own agents that work simultaneously.

“When a task is significant enough, it splits into separate sub-agents operating in parallel within distinct environments,” Zuckerberg elaborated. “Your working copy remains unchanged. During testing, it successfully created six features for a game concurrently without any conflicts.”

This initiative aims to place Meta in a competitive and cost-effective position alongside AI research counterparts like OpenAI with its Codex coding assistant, and Anthropic’s Claude Code.

“We believe that for numerous workflows and varied use cases, this could offer exceptional value, especially regarding cost,” remarked Alexandr Wang, Meta’s AI lead overseeing Meta Superintelligence Labs, during an interview with the Wall Street Journal.

Meta has been actively striving to enhance its AI footprint by investing heavily in development. In June, it expanded its focus beyond traditional advertising to enter the enterprise AI space with a tool designed for customer service and support.