Amjad Masad's company says users set the goal, success criteria and spending limit, then agents plan, build, test and iterate until the judge signs off or the money runs out.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Replit launched Missions on November 17th, a product that assigns a swarm of AI agents to a defined outcome and stops them when the work passes review, needs human input, or hits a spending limit the user set in advance.
Video from the original post on X.…