Brilliant accelerates feedback loops and eliminates CI friction with RWX

About

Brilliant is a corporate gifting and branded merchandise company, helping teams design, produce, and deliver branded gifts at a global scale. Founded in 2015, the company offers an end-to-end service that combines product curation, campaign design, gifting automation, inventory tracking, and worldwide fulfillment for enterprises including Anthropic, Airbnb, and Spotify.

Challenge

Growing test volume pushed CI beyond Brilliant's target feedback window

Brilliant automates the workflows behind global gifting programs, making reliable software delivery central to how the company serves customers. But as Brilliant expanded headcount alongside curated investments in AI-assisted development, its test suite was growing faster than ever, stretching build times to 25 minutes on average.

David Pick, CTO at Brilliant, watched build times stretch firsthand. Engineers would lose hours a day waiting for pull requests to clear before they could merge and move to the next task. If a failure arrived half an hour later, the engineer had to recover context, correct the issue, and restart the cycle. The same delay lengthened the path to validating and deploying an urgent production fix.

“I like to keep build times under 10 minutes because of the way feedback cycles work,” David explains. “At 20 to 25 minutes, engineers are sitting there waiting, switching context, and then coming back when a test needs attention.”

Brilliant had already built custom caching logic in GitHub Actions, which enabled the team to reuse some completed work. But engineers still had to create and maintain that behavior themselves. David had used other CI providers, but they were exorbitantly expensive and still required his team to recreate much of the custom logic. He needed a platform with cross-build reuse as standard, so valid task outputs could be restored automatically whenever inputs remained unchanged.

David was already familiar with RWX through its service, Captain, which he had connected to Brilliant's existing pipeline to detect and remedy flaky tests. Builds that would previously have failed began passing, while Captain kept the underlying flaky tests visible for repair. That early success, combined with RWX's fundamentally different caching strategy and cost-effectiveness, gave David the confidence to expand into CI.

Solution

RWX replaces isolated builds with a reusable execution graph

In David's words, “Onboarding with RWX was super easy and smooth.” He shared Brilliant's existing GitHub Actions configuration and the RWX team translated it into a working pipeline within two weeks. Brilliant didn't have to learn a new system from scratch to get running, and engineers kept the same pull-request and review workflow they already knew.

Now, David's build and test pipelines run as a directed acyclic graph of tasks and dependencies instead of a sequence of isolated jobs. Independent tasks run concurrently, while RWX's content-based cache checks whether each task's inputs have changed. When they haven't, RWX restores the previous output, including reusable setup and package-download work.

That shift directly addresses the feedback-loop problem David was trying to prevent. CI now responds to what actually changed in the code, instead of running the same fixed sequence and making every pull request wait on it. For David's team, that means the pipeline execution time went from 20 to 25 minutes to about three. And when a production fix is urgent, the same shorter loop gets it validated and deployed sooner.

Just as important, reuse is inherent to RWX rather than another layer that David's team has to maintain. David no longer needs to hand-build and port caching logic between providers. RWX's execution model handles that reuse automatically, cutting both runtime and the engineering attention required to keep CI efficient as the test suite grows.

For David, the value isn't just a faster version of the old pipeline. The amount of CI work now scales with what actually changed, not with the size of the whole workflow. That gives Brilliant room to add tests and strengthen validation without adding the same penalty to every engineer's feedback loop.

Results

Fast, dependable CI that keeps pace with platform growth

With RWX, Brilliant unlocked a dependable validation layer built to absorb continued test growth without imposing the same full-pipeline penalty on every change. Engineers now receive feedback quickly, which keeps development moving, while David can rely on the platform to handle more build and test work without commanding internal bandwidth.

Since partnering with RWX, Brilliant has cut CI/CD pipeline execution time by 85% with no increase in CI costs compared to GitHub Actions.

As Brilliant grows and AI-assisted development adds more tests to every build, David trusts RWX's team and their caching model to keep pace without pulling his engineers back into CI maintenance.