Replit’s agent swarms built an app monitoring platform in 2 days on a $3K budget. Now, it’s offering them to customers.
Agent swarms are all the rage, as companies set about finding ways to coordinate multiple AI agents to tackle complex The post Replit’s agent swarms built an app monitoring platform in 2 days on a $3K budget. Now, it’s offering them to customers. appeared first on The New Stack .
Replit's Missions product enables users to set objectives, budgets, and approval requirements, and then deploy multiple AI agents to achieve those goals. Users describe their objective in plain English, and agents determine the best way to approach the problem, adapting their methods as needed. Replit president Michele Catasta says this design allows AI systems more flexibility in finding effective solutions, rather than users having to map out every step in advance.
Missions agents operate in isolated cloud environments, each with access to development tools, integrations, and infrastructure. Users set a spending limit, and Replit checks costs at every step, blocking further use once the limit is reached. Missions was used to build Replit's Radar application monitoring platform, completing it in two days with a $3,000 budget.
The system identified a database fault, raised an alert in 38 seconds, and processed 26 million activity records and 4.4 million log entries without losing data in a three-hour test. Missions also improved Replit Slides, the company's AI-powered presentation creation tool, by up to 19 times using 32-agent swarms. Replit plans to release Radar to the public under an open-source license after further reliability improvements and a code audit.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.