October 26, 2026
Startups
How Greptile uses GPT‑6 Astra to test user workflows, Daybreak to harden security, and Codex to reduce rework.
Company size: Startup
Region: North America
Industry: Technology
Products: API, Codex
- Testing applications at scale with GPT-6 Astra
- Strengthening customers’ defenses with Daybreak
- Reducing bugs and rework with Codex
- Looking ahead to faster development with Ultrafast
Greptile builds AI agents that review and test code changes with full context across a codebase. “The core of our business is reducing bugs and making software more reliable around the world,” says co-founder and CEO Daksh Gupta.
For Daksh, that mission requires not just reading code, but testing an application’s runtime behavior—and GPT‑6 Astra’s computer use capabilities provide Greptile with a path toward automated testing at scale.
“For the first time ever, computer use is at a point where you can simulate and automate what a real user would do to try and break an application, giving massive enterprises a lot more confidence to ship code.”
—Daksh Gupta, Co-founder and CEO, Greptile
Testing applications at scale with GPT‑6 Astra
Validating a pull request means verifying whether code logic is sound, matches developer intent, remains secure, and avoids breaking existing functionality. For some of these checks, you actually have to run the application, as reading the code alone may not reveal issues.
Daksh points to the example of a button disappearing when browsing a website on an iPhone. With GPT‑6 Astra, Greptile can spin up a dev server for a pull request, resize the browser to an iPhone-sized viewport, and simulate a user navigating the application. The model’s visual understanding allows it to inspect the screen and catch the hidden button.
“Given how fast we’re shipping code with these models, having a QA firm test all your changes is wildly impractical,” he says. With GPT‑6 Astra, Greptile can automate this kind of application testing at scale.
Strengthening customers’ defenses with Daybreak
Daksh notes that more capable models also make it easier for attackers to exploit application vulnerabilities, putting companies at risk—even those that haven’t traditionally been typical targets.
“We want to make it insanely easy for even startups to secure their systems,” says Daksh. “Daybreak lets us do that in a very safe way, ensuring our customers have defenses that are more sophisticated than the attacks they face.”
Reducing bugs and rework with Codex
Greptile’s Knowledge Base indexes a company’s codebase and documents how it works in a format that both humans and agents can understand. The company serves these documents to Codex via Model Context Protocol (MCP) servers.
According to Daksh, this enables Codex to write code that aligns with a customer’s existing patterns, and thousands of customers have already connected Codex to Greptile’s Knowledge Base.
To measure success, the team tracks how often AI-generated code gets reverted. “When we give the agent context, code reverts drop significantly, saving customers both time and inference compute,” says Daksh. Customers can invest those saved resources into new initiatives rather than repeatedly reworking the same changes.
Greptile also uses Codex to build its own software and agents. “We’re massive users of Codex,” says Daksh. “Our usage is through the roof across every function, especially engineering.”
Looking ahead to faster development with Ultrafast
Daksh explains that for Greptile, the only real bottleneck is how fast the team can turn ideas into working software. He notes that model intelligence is already more than sufficient for many tasks, so he sees Ultrafast as a powerful lever to speed up development.
“What excites me most is that Ultrafast can be the lever that lets us move as fast as we can have good ideas.”
—Daksh Gupta, Co-founder and CEO, Greptile
His long-term goal is to make modifying and shipping critical enterprise software as effortless as launching a personal side project, and OpenAI is helping Greptile build the validation capabilities needed to get there.