Jane is a health and wellness practice management platform serving practitioners and their patients with seamless scheduling, billing, and telehealth solutions.
Jane's Delivery Engineering team is responsible for ensuring that Jane's development pipeline is efficient, reliable, and cost-effective for over 250 developers spread across 35 engineering teams. By adopting Depot, Jane significantly decreased their build times, cut GitHub Actions spending in half, and increased engineering throughput by 25%. We spoke with Alonso Suarez, a Staff DevOps Engineer on the Delivery Engineering team, to learn how Depot became one of Jane's most impactful infrastructure investments.
The challenge
Many of our developers became incredibly vocal that the pipeline wasn't great. There were a lot of struggles that deeply impacted our team, and people just weren't excited to work with it.
Prior to adopting Depot, Jane's engineering teams experienced slow and unreliable pipelines that significantly impacted productivity as well as the team's happiness with their work.
One issue was that Docker image downloads were slow – really slow. "We were spending most of our time just downloading Docker images," Alonso told us. To remedy this, they re-architected their CI to promote a single Docker image (instead of building multiple images on every parallel job), but this surfaced new challenges. The first was their choice of registries for storing their built images: They chose DockerHub and GitHub Container Registry, but both services suffered from frequent instability, leading to significant frustration for the Jane team.
Another issue the team faced was that the delivery pipeline had only a 40% success rate and required a manual retry. "That meant that for 6 in every 10 runs, you had to rerun the build," Alonso said. "And after you clicked rerun, it would just pass," he added, highlighting the unpredictability and lack of control for developers. "That is the worst developer experience ever."
To address the random failure problem, Jane's leadership launched a company-wide quality quarter, shifting priorities across all engineering teams to fix flaky tests and stabilize their pipelines. This effort brought their build pipeline success rates up to 90%, but they still needed a faster and more cost-effective CI/CD solution, which highlighted another problem: an over-abundance of vendor solutions.
The delivery team was managing three different CI systems — AWS CodeBuild, CircleCI, and GitHub Actions — until they made the decision to consolidate onto GitHub Actions. The migration was a massive effort spanning four months, and in the end, came with some sticker shock: Even though they could now run fewer jobs, the cost of GitHub Actions ended up being even higher than what they'd been paying before.
The solution
After learning about Depot, Alonso's team recognized it as a potential solution for Jane's core problems around build speed and infrastructure cost. They initially tested Depot's GitHub Actions runners and build acceleration tools, evaluating both cost and performance, and a few key benefits became clear:
- Faster builds: Depot UltraRunners provided a massive performance boost with a simple configuration change.
- Eliminating redundant rebuilds: Jane had 80 parallel runners, each of which previously rebuilt the app from scratch. Depot optimized their process to build images just once, and then reuse that image across builds.
- Proactive support: Depot's team worked closely with Jane, even writing custom GitHub Actions to further optimize Docker image downloads.
Alonso highlighted both the ease of implementing Depot, and the immediacy of results. "We observed a 20% decrease in build time just by changing one line," he said. "That's massive."
Alonso also highlighted Depot's hands-on technical support. He noted that at one point in their implementation, Jacob – Depot co-founder and CTO – "created an action to use a snapshotter to download the Docker image in parallel," further increasing their build speed gains.
And unlike traditional vendors with slow, ticket-based responses, Depot provided real-time support via Slack, helping Jane optimize their CI/CD workflows in real time. "I feel like Depot is an embedded team at Jane," Alonso said. "I now have experts like Kyle, Jacob, and Chris Goller on Slack Connect, and the SLA is like, I don't know – 35 seconds? Insane."
The measurable impact
Last year, the most impactful project we did for Engineering at Jane was migrating to Depot. It was one week's effort and one month's lead time. It wasn't a high-cost implementation, yet it was the most impactful.
Jane's migration to Depot delivered tangible benefits in build speed, throughput, and cost savings:
- 2.4x faster key end-to-end test jobs
- 25% increase in build throughput
- 55% reduction in GitHub Actions spend
Reflecting on Depot's impact on Jane, Alonso shared, "My director came to me and he said, 'Should we host this ourselves?'", to which Alonso responded, "Absolutely not! These guys have optimized network runners, optimized disk runners, optimized Mac runners. All of that is at my disposal – I don't want to go and build my own pipeline."
Looking ahead
With Depot now fully integrated into their development workflow, Jane has successfully transformed their CI/CD pipeline into something that is fast, cost-effective, and frustration-free for the developers who use it day in and day out.