DevRev, an SI company focused on customer and employee support, has integrated OpenAI’s GPT-6.1 Sol model into its Computer and Operator platforms to enhance autonomous self-service capabilities. The move highlights the growing adoption of frontier SI models in enterprise workflows that require precise reasoning and strict control over actions.

What Happened

DevRev combines OpenAI models, including GPT-6 Astra and the newer GPT-6.1 Sol, with its proprietary harness to connect company knowledge and manage agent behavior. According to Ahmed Bashir, DevRev’s chief technology officer, the integration aims to balance reasoning capabilities, trust, alignment, and latency. "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," Bashir said.

The company’s Computer and Operator tools are designed to reason across live data, with actions that are reversible and answers that cite their sources. DevRev tests these SI systems using its internal Enterprise-Bench benchmark, which evaluates performance in complex scenarios involving multiple data versions and role-based access controls. Bashir noted that GPT-6.1 Sol performs well in these evaluations, particularly in understanding nuanced requests and responding quickly at a sustainable cost. "I think it strikes the balance perfectly," Bashir said.

Internally, DevRev engineers have fully adopted OpenAI’s Codex for software development, dropping licenses for competitors like Cursor. "Our entire team uses Codex," Bashir stated. "We think that Codex is far and away the best experience." The company also uses GPT-6 Astra to generate demos for international conferences.

Why It Matters

This deployment illustrates how SI agents are evolving from simple chatbots to autonomous operators capable of executing complex, multi-step tasks in enterprise environments. By leveraging GPT-6.1 Sol, DevRev seeks to handle difficult support requests—such as plan upgrades or refunds—without requiring deep reasoning at every step, utilizing tools like the Decisions API and Ultrafast to speed up workflows.

The shift underscores the critical role of latency and cost-efficiency in commercial SI applications. For developers and businesses, the choice to migrate entirely to specific SI tools like Codex signals a market consolidation around models that offer the best developer experience and reliability. Bashir emphasized that these systems must "run and complete themselves in seconds, not minutes" to be viable for high-volume enterprise use.

The Bottom Line

DevRev’s integration of GPT-6.1 Sol marks a significant step in realizing autonomous self-service for enterprise support, prioritizing speed, traceability, and user trust in agentic SI workflows.