DevRev Leverages GPT-6.1 Sol for Autonomous Self-Service

·

DevRev has successfully integrated the GPT-6.1 Sol model into its Computer and Operator platform, enabling autonomous self-service capabilities for enterprise service teams. This integration balances reasoning, speed, and cost, making it an attractive solution for businesses looking to streamline their operations.

The Computer and Operator platform was designed by DevRev to help service and operations teams handle even the most complex customer and employee requests with ease. The system is built on live data and allows actions to be reversible, while answers are linked back to their sources, according to Ahmed Bashir, DevRev’s chief technology officer.

DevRev combines its own proprietary models with OpenAI’s GPT-6 Astra and GPT-6.1 Sol to provide reasoning and reliability at scale. The company sees these models as a key component in delivering fast and cost-effective customer service. ‘It’s about striking the right balance between reasoning capabilities, trustworthiness, alignment, and latency,’ Ahmed explains.

The OpenAI models power the reasoning, planning, and tool selection behind both Computer and Operator. In DevRev’s Agent Studio, users can define how they want a tool to behave in conversation with AI and specify parts of its behavior using fixed rules. This allows for greater control over the agent’s actions and ensures that it is aligned with company policies.

DevRev tests its models together with software that provides context, enforces permissions, and controls actions. The Enterprise-Bench benchmark evaluates how well they perform in scenarios where data has multiple versions, information is scattered across different systems, and access depends on a person’s role. Ahmed reports that GPT-6.1 Sol performs exceptionally well in these evaluations.

The company’s engineers have made a deliberate choice to use Codex as their primary coding tool. This decision grew organically after they explored various options and found Codex to be the best fit for their needs. By the time of our interview, DevRev had fully transitioned to Codex and discontinued its Cursor licenses. Ahmed attributes this shift to his team’s desire for the most effective tools available.

DevRev’s sales engineers also utilize GPT-6 Astra to create demos for international conferences. When a customer or employee requests assistance, Operator may need to understand the issue, take approved action, or route it to Computer to prepare the right person to act. Ahmed sees opportunities for the Decisions API and Ultrafast to enhance these workflows by allowing them to complete quickly without requiring deep reasoning at every step.

Ahmed wants developers to be able to describe more of their logic conversationally and choose the right type of intelligence for each task. He emphasizes that decisions must run and complete themselves in seconds, not minutes. DevRev’s next ambition is for Operator to handle longer, more demanding requests while maintaining action traceability, permissioning, and reversibility.

DevRev envisions a future where Computer and Operator can tackle complex customer and employee requests with ease, leveraging company knowledge and controls to act on behalf of individuals. Ahmed sees familiar support tasks as just the beginning, aiming to realize the full promise of self-service through OpenAI’s frontier intelligence.

Discussion

No messages yet. Be the first to write.