Replit Sends Support Cases Through AI Agents Before Human Review

Replit has put AI agents at the start of its support process, where they gather technical context before employees decide how to handle incoming requests. Replit CEO and co-founder Amjad Masad described the workflow as an example of how people and software may share work as AI agents take on routine tasks.

“Our company entirely changed how we operate, and I would say the closest analogy is a ‘self-driving company,’” Masad said at a San Francisco conference Thursday. He used the phrase to describe a company where agents handle much of the preliminary work while people contribute judgment and taste.

When a support request arrives, an agent checks relevant GitHub issues, Replit’s codebase and Linear tickets. It passes that information to an employee, who decides how to respond. If the employee determines that a fix is needed, they can launch another agent to help prepare a patch.

Masad said the arrangement leaves a person involved in deciding what to do with a request. “You need human judgment; you need taste,” he said. The company’s agents handle much of the work around a case, while employees make decisions about responses and fixes.

Masad said Replit staff now spend more time choosing what to work on, discussing ideas and planning software architecture. He linked the change to agents becoming more reliable earlier this year, prompting the team to consider how AI had changed its own workplace.

He also said conversations with other businesses focus on how AI might help employees become more productive and find more creative work. Businesses use Replit’s AI coding tools to develop and launch their own software, while Masad presents the company’s internal support process as one way agents can take on routine tasks alongside human workers.