DevRev this week updated its Computer platform to enable teams of end users to collaborate using the same artificial intelligence (AI) agents to automate tasks.
Computer, by DevRev, is based on a framework that enables the AI agent to understand how data is organized, how an organization is structured and how people interact with that data. That approach ensures the output provided by the AI agent is both more accurate and relevant, says DevRev CTO Ahmed Bashir. Rather than making a guess, the AI agent will generate the same answer to the same question every time because it has been grounded on data that has been validated, he adds.
The platform has been specifically designed for enterprises that need to understand and audit how an AI agent is performing a task and make sure the outcome has been approved by a human, says Bashir. Additionally, the framework reduces the number of tokens an AI agent would otherwise consume by reducing the amount of data that would need to be added to a context window, he notes.
The latest release of the AI agent adds a Multiplayer AI capability that enables multiple users to share a live Computer session via an application. “It creates a team experience,” says Bashir.
There is also now an Agent Studio tool that gives teams the ability to build, test in a sandbox environment, and safely deploy them across connected systems. Connectors to Gmail, Outlook, Slack, Notion, Google Drive, Jira, Microsoft OneDrive and SharePoint are available, or the AI agent can be integrated with any tool that supports the Model Context Protocol (MCP).
Rather than being a question-answering tool, the platform is evolving into a content-producing system that enables teams to build on each other’s work and correct reasoning in real time, says Bashir. From within that application, teams can now generate complete, fully branded and formatted work artifacts grounded in real business data using reusable skills that have been embedded into an agentic AI workflow, he adds.
Every action runs under individual user permissions, not a shared account. Every step is traceable, auditable, and reversible. If an agent makes a mistake, it can be rolled back. Teams can create reusable workflows based on natural language queries that invoke the Text2SQL language.
The overall goal is to enable enterprises to safely add AI agents to workflows in a way that ensures they have full control, says Bashir.
Existing customers include BILL, HDFC Bank, FAME, Pebl and Uniphore. The latter two organizations are now using the platform to resolve 85% of support tickets without any human involvement, DevRev claims. In total, more than 250 organizations are using the AI platform in production, with more than 1,000 users onboarded since the initial launch last September.
It’s still early days so far as adoption of AI agents in the enterprise is concerned. The one thing that is certain, however, is that the more care that is taken to structure the data an AI agent can access, the better the overall outcome will be for all concerned.
