DevRev builds self-service agents on GPT-6.1 Sol and GPT-6 Astra

Agents built for hard requests

DevRev created Computer and Operator to help service and operations teams finish even the hardest customer and employee requests. According to Bashir, both are designed to reason across live data, actions are reversible, and answers point back to their sources.

DevRev pairs OpenAI models, including GPT-6 Astra and GPT-6.1 Sol, with its own harness. The harness connects company knowledge and gives teams control over which actions an agent can take. OpenAI models power the reasoning, planning and tool selection behind both agents.

Bashir described what DevRev needs from a model: "It's a combination of being able to have the right reasoning capabilities, being able to be trusted, highly aligned, and most importantly, it has to work from a latency perspective."

Testing models with company context

In DevRev's Agent Studio, users can describe in a conversation with AI how they want a tool to behave, and they can define parts of its behavior with fixed rules.

DevRev tests the models together with the software that supplies company context, enforces permissions and controls actions. Its Enterprise-Bench benchmark checks how well they work when data has multiple versions, information sits in different systems, and access depends on a person's role. Bashir says GPT-6.1 Sol performs well in those evaluations.

An agent talking to a customer or employee must grasp the nuances of a request and answer quickly, at a cost the business can sustain. "I think it strikes the balance perfectly," Bashir says.

Codex inside DevRev

DevRev also uses OpenAI tools internally. "Our entire team uses Codex," Bashir says. Adoption grew organically after engineers explored several coding tools, and by the time of the interview the company had moved fully to Codex and dropped its Cursor licenses.

Bashir says this reflects what developers want: "Developers are really just looking for the best tool, and you can't fool them." He adds that DevRev's engineers were unequivocal about continuing with Codex. Sales engineers use GPT-6 Astra to create demos for the company's booths at international conferences.

Faster workflows and the goal of self-service

A request may require Operator to understand the issue, take approved action, or route to Computer to prepare the right person to act. Bashir sees opportunities for the Decisions API and Ultrafast to help those workflows finish quickly, without deep reasoning at every step. Some decisions are hard and expensive to define with fixed rules, especially when the underlying information keeps changing, so he wants builders to describe more of that logic conversationally and pick the right kind of intelligence for each task.

"These things have to be able to run and complete themselves in seconds, not minutes," he says.

DevRev's next ambition is for Operator to handle longer, more demanding requests while keeping actions traceable, permissioned and reversible. Bashir treats familiar support tasks such as finding a knowledge-base answer, upgrading a plan or getting a refund as only the beginning. "We want to really realize that promise of self-service," he says.

What it means

The case shows how DevRev is positioning agents for support and operations work: model reasoning combined with company knowledge, permissions and reversible actions. According to OpenAI, DevRev is building toward Computer and Operator resolving more complex requests and acting on a person's behalf. The source gives no pricing, availability or benchmark figures.