Summary
OpenAI’s customer spotlight series features Ramp’s head of AI describing how the finance company has been using GPT-6 Astra’s computer-use capabilities to speed up internal engineering and automate parts of its product. He compares working with Astra to delegating to a trusted coworker: instead of writing detailed step-by-step instructions, he can hand off a vague, high-level feature request and check back later.
He walks through a specific example — asking Astra to implement API-key-level routing strategy controls with a single, minimal prompt — and shows how the model correctly located the right settings in the product, made the change, and validated the result on both desktop and mobile views without being told to. The segment highlights computer use as a growing method for verifying low-risk code changes without manual QA.
Looking ahead, he explains that Ramp plans to apply Astra’s computer-use abilities to one of its core products: automatically fetching customer receipts from merchant websites, a task that currently requires a lot of manual prompting to handle the long tail of different merchant site layouts. The video offers a real-world look at how an enterprise fintech company is integrating agentic, screen-interacting AI into its engineering and product operations.
📺 Source: OpenAI · Published September 21, 2026
🏷️ Format: Workflow Case Study
