Tropic slashed CI runtimes by 67% with RWX
Tropic migrated from a bloated GitHub Actions workflow to RWX, cutting CI runtimes by 3x while accelerating the path from PR to deployment.
"RWX was built by engineers for engineers, and it shows in how fast and hands-off CI feels day to day. Now, our runtimes are down to 10 minutes, and shipping with confidence is no longer the exception; it's the norm."
#About
Tropic is a software procurement platform designed to help organizations gain complete visibility and leverage over their business spend and make smarter purchasing decisions that save time and money. Recognized by G2 as a global leader in SaaS Spend Management and Procurement Orchestration, Tropic is trusted by over 500 organizations, including Strava, Greenhouse, Mozilla, Olipop, and more.
#Challenge
#Tropic's CI feedback loop was drifting from guardrail to bottleneck
Like many leading B2B SaaS companies, Tropic treats CI as the crucial guardrail that keeps daily releases safe as its product evolves. However, as the company's codebase and team size continued to expand, that same guardrail quickly began to impede development velocity rather than accelerate it.
Previously, Pedro Coronel and Luke Scoates, Principal Engineer at Tropic, were running CI through GitHub Actions with hosted runners, executing several parallel test processes on every PR. As the testing suite grew, runtimes climbed from 10 to 30 minutes for a full run. CI stopped being a quick verification step and began commandeering how the team planned, reviewed, and shipped changes. Instead of tightening the loop between coding and deployment, CI became the slowest step in the workflow.
With even routine changes increasingly being stalled, Pedro's team tried to buy back time by scaling runner capacity and increasing parallelism. However, with each improvement came a corresponding increase in ongoing CI spend that wasn't sustainable. Tightening slow specs and removing redundant tests also proved ineffective, as runtimes continued drifting upward as the product grew.
Meanwhile, intermittent failures added avoidable churn: a run could execute thousands of tests and fail on one or two, and without a targeted recovery path, engineers often had to rerun the entire suite just to confirm whether the failure reflected a real regression or transient noise. With CI running repeatedly throughout the day, those delays compounded quickly. Rather than throwing spend or headcount at the issue, Pedro and Luke sought an established CI platform that would eliminate this validation bottleneck and empower their team to ship with speed and confidence.
Pedro discovered RWX, which gave Tropic a credible path to evaluate an alternative. After RWX ran Tropic's CI pipelines alongside the existing GitHub Actions implementation, Pedro saw measurable runtime reductions, giving him the confidence in a broader cutover.
"It doesn't take long for slow CI to block critical engineering work. As runtimes nearly tripled in duration, we knew our existing workflows wouldn't align with our growth trajectory."
#Solution
#A modern CI platform that ensures speed, stability, and code quality
Implementing RWX gave Tropic a faster, more predictable CI loop without requiring a drawn-out migration or increasing runner capacity costs.
From day one, the RWX team worked with Tropic's engineering leadership to map the existing CI flow, define the migration path, and build the scripts for a clean cutover. They then ran a live walkthrough to show Tropic's engineering organization the capabilities available on RWX that weren't possible on GitHub Actions and adopt day-to-day workflows, even making targeted codebase updates to harden brittle tests and reduce avoidable churn.
Because RWX runs CI as a directed acyclic graph (DAG) of discrete, cacheable tasks, the platform enables true parallelization of independent work, rerunning only what fails, and reusing prior outputs when nothing relevant has changed. With this foundation in place, Pedro credits RWX with improving Tropic's CI workflow in several concrete ways:
#Faster runtimes without escalating CI spend
RWX helped shorten Tropic's CI runtimes by enabling independent steps to run in parallel instead of queueing behind a single job. Because each task is scheduled with its own CPU and memory allocation, Tropic improved wall-clock time without having to keep scaling runner capacity to stay fast.
That translated into faster PR iteration without turning runtime targets into an ever-increasing infrastructure bill. Instead of paying for speed by adding more runners, Tropic keeps runtimes low while maintaining CI infrastructure growth.
#Removing flaky tests as a roadblock to deployment
RWX reduced the cost of failed runs by enabling Tropic to rerun only the failed task rather than restarting the entire pipeline. Because completed tasks are cached and reused, the team can skip work that doesn't need to be recomputed, turning a failure into a targeted rerun rather than a full reset.
#Accelerating merges and deployments with caching
With that same cache-and-reuse foundation in place, Tropic shortened post-merge validation by reusing verified work when inputs hadn't changed, so checks that already passed in PRs don't need to run again after the merge. This tightens the path from merge to deployment by turning post-merge verification from "run it all again" into "reuse what's already proven."
#Managing intermittent failures with visibility and quarantine workflows
With RWX, Tropic surfaces failure patterns through Captain by aggregating test results across runs and tracking which tests failed intermittently over time, making it easier to isolate recurring offenders and prioritize fixes.
When a recurring offender emerges, Pedro's team quarantines the test so it still runs and stays tracked, but no longer blocks builds while a fix is in progress. "It's almost a one-click workflow," Pedro notes. "You can unblock the team immediately while you work on a fix, without letting one test stall everyone else."
"RWX got our CI runtimes from 30 minutes to around 10, basically overnight. At roughly 50 runs a day, the time and cost savings add up fast."
#Results
#Faster releases, reclaimed bandwidth, and a CI foundation that scales with product growth
With RWX, Tropic transformed CI into a fast, dependable control point that supports engineering output rather than bottlenecking it. Now, Pedro's team continuously ships improvements to Tropic's procurement workflows without waiting on the CI pipeline to catch up.
The results since the partnership began:
Looking ahead, Pedro and Luke see RWX as infrastructure Tropic can rely on as the codebase and test suite continue to grow – keeping CI fast, trustworthy, and low-friction.
"The RWX team moves quickly, and we can't wait to see what solutions they have in the pipeline. We knew early on that they would be a valuable long-term partner for Tropic."