Adam Gee, VP of engineering at Rubrik, didn’t change his mind about AI after a demo or a roadmap. It happened while watching an agent work end to end on Rubrik’s real codebase. He gave Auggie an idea—not a snippet, not a carefully engineered prompt—and watched it get to work. It inspected the codebase, built a plan, made coordinated approved changes, ran unit tests, fixed failures, and re-ran them until the system passed. “That was a real game-changer moment for me.”
What stood out wasn’t that Augment wrote code quickly. It was that the agent could reason about a large, messy, production system and carry work through to completion without hand-holding.
That moment became the reference point for everything that followed—how Rubrik evaluated tools, how engineers adopted them, and how AI spread beyond individual productivity into a new way of working.
Where others stalled – and Augment came to life
Rubrik had already been down the AI coding path.
Other tools came along and engineers tried them. In a small repo, it helped. In Rubrik’s 12‑year‑old monolithic codebase, it didn’t.
“With a huge monolithic codebase, we found that other agents weren’t really adding value.” And engineers are pragmatic. “If a tool doesn’t add value, engineers drop it very quickly.”
That experience set a baseline—and made adoption harder the next time around.
So when Rubrik evaluated Augment, expectations were low. What changed was context.
Augment’s Context Engine indexed the entire codebase—multiple product lines, languages, and years of accumulated complexity. A cross‑org team evaluated tools across experience levels. Augment won.
Trust and security weren’t optional
Rubrik is a Security and AI operations company. That changes the bar.
Two things mattered in their evaluation:
- Deep context across a large codebase
- Enterprise‑grade security and governance
“Our internal processes are very security‑focused, and Augment took those concerns seriously.”
Rubrik’s InfoSec team pushed hard: controls on tool calls, policy enforcement, governance over what agents could and couldn’t do. Augment responded and iterated.
“We worked as partners to make sure we were doing AI responsibly.”
That trust unlocked broader adoption.
Beyond code: attacking the other 80%
Adam points out a stat that reshaped Rubrik’s strategy:
“Only about 20% of a developer’s time is spent writing code.”
The real opportunity was the other 80%—triage, debugging, release management, operational glue work. Rubrik started experimenting with agent personas.
“We said: ‘Auggie, become a triager.’”
Agents scraped logs, correlated signals across systems, and reduced hours of manual toil.
Soon, teams began contributing their own personas.
“We now have around 100 prompts created organically across the company.”
A multipurpose agent, grounded in real context
Even non‑engineering roles leaned in. Release managers automated Jira scraping and coordination. Support teams installed IDEs just to access Augment and understand stack traces.
“With the right context and tools, Auggie became a triager, a reviewer, a release coordinator, and a support assistant. Not all use cases you’d expect from a coding agent—but it works because of context. It opens up use cases for us that just weren’t possible before.”
The takeaway
Rubrik’s journey with Augment didn’t start with a mandate. It started with an engineer watching an agent do real work—and trusting the result.
From there, adoption followed naturally. Not because engineers were told to use it, but because once they saw it, they didn’t want to give it up. And for a security and AI first company with a complex codebase, that trust made all the difference.
About Rubrik
Rubrik is a cybersecurity company that secures the world's data with Zero Trust Data Security™. As a 12-year-old company with multiple product lines and platforms, Rubrik operates a large, complex monolithic codebase while maintaining security-first internal processes