Overview
Willow Wealth's engineering org runs lean: around 20 engineers, a small platform team supporting product squads, trunk-based development, and multiple deploys a day. That cadence only works if developers can validate changes quickly, and trunk-based development doesn't lower that bar — it raises it — which is exactly where their homegrown ephemeral-environment system started to fail them.
Environment creation had crept past 40 minutes and stopped being reliable enough to trust. Today, with Bunnyshell, the same environment comes up in under five minutes — and pre-merge validation now happens on that environment itself, instead of waiting for a staging slot.
Company Background
Willow Wealth runs its own cloud infrastructure on Kubernetes, deploying services via Helm, with GitHub Actions as primary CI. Because developers own quality end-to-end — there's no separate testing team — on-demand, prod-like environments aren't a nice-to-have; they're the difference between validating a change properly and shipping on hope.
Before Bunnyshell, the platform team had built an internal solution meant to give every PR its own preview environment, retire the shared-staging bottleneck, and give developers reproducible environments on demand. Functionally, it mostly worked. Operationally, it became a platform nobody had budgeted to maintain.
Slowed Down by a Platform Nobody Owned
Day to day, environment creation took 40+ minutes and wasn't reliable enough to trust. Builds frequently failed mid-deploy, so developers got into the habit of spinning up four or five environments in parallel, hoping one would land — burning cloud spend on environments that mostly died before anyone used them.
- PR reviews stalled because reviewers couldn't click around the change.
- Pre-merge validation piled up at the end of every sprint.
- Releases went out with less confidence, and sometimes slipped.
- Developers routed around the platform to shared staging — recreating the exact queueing problem it was built to solve.
Finding the Right Fit
The old system's real failure wasn't just speed — it was that no one owned it, and every hour spent nursing it back to health was an hour not spent on work the business was actually paying for. So when the platform team went looking for a replacement, the requirements were strict: single-tenant deployment in their own cloud account, native Helm support, clean integration with GitHub Actions without a pipeline rewrite, full API/CLI parity, and enough customization to accommodate legacy components that don't fit a cookie-cutter template.
They evaluated Qovery and Release.com as the two main alternatives. Both were credible products, but neither cleared the bar on the combination of native Helm support, the single-tenant, in-their-own-cloud deployment model, and the customization their legacy components needed. Cost was the other differentiator: the pricing models they were quoted would have scaled badly as the org grew — non-starters for a company actively adding engineers and services.
Accelerating Delivery with Bunnyshell
Speed and reliability made Bunnyshell click first — a real Helm-based environment came up in their cluster in under five minutes and stayed up. The bunnyshell.yaml model mapped cleanly onto how their services were already organized, and the pricing didn't penalize them for adding engineers or services.
Today, a developer kicks off an environment from the Bunnyshell UI or CLI to validate a change. Bunnyshell provisions it against their own Kubernetes cluster in under five minutes, and the environment URL surfaces in the dashboard for the developer and anyone they share it with. From there, branch-tracking takes over: as CI produces new builds, Bunnyshell deploys them into the environment automatically via its API, so the environment never drifts out of sync with the branch — without Willow Wealth having to move its CI build into Bunnyshell at all. GitHub Actions stayed exactly as it was.
A typical environment now spans around 40 components pulled from a deliberately heterogeneous source layout — a single frontend monorepo alongside a long tail of independently repo'd backend microservices, each deployed via Helm, with containerized databases provisioned alongside them for that environment's own use. Bunnyshell handles that mixed layout without tripping up, adding up to close to a full slice of the production stack per environment.
Faster, More Confident Releases
The jump from 40+ minutes to under five didn't just save time — it changed behavior:
- Full-stack testing became the default, not the exception — developers spin one up on demand, not as an automatic step tacked onto every PR.
- Pre-merge validation moved onto the PR environment itself, instead of waiting for staging.
- Product and design started previewing in-flight features as real interactive environments, not Figma mockups.
- Developers began running multiple parallel experiments without worrying about cost or breakage.
- Bug hunts became a one-shot job — a fresh environment off the exact commit, instead of blocking shared staging for hours.
- Engineers also connect directly into a Bunnyshell environment (Telepresence-style), testing changes against the full stack without a CI round-trip.
Willow Wealth now ships more often and catches more issues before merge, because developers, reviewers, and product are all pointing at the same environment.
Deploy Queueing: A Feature Born From Feedback
Under the old homegrown system, too many environments deploying in parallel could overload the cluster and cascade into timeouts. Bunnyshell shipped deploy queueing directly in response to that feedback, letting Willow Wealth scale parallel deployments without the risk — and eliminating a class of incidents they used to deal with regularly.
Willow Wealth on Working With Bunnyshell
Onboarding ran against Willow Wealth's actual stack from day one, not a sample app. What stood out most, according to the team, was how hands-on the Bunnyshell engineers were — debugging in real time, understanding their setup, and never making them feel like a support ticket.
What's Next for Willow Wealth
Looking ahead, Willow Wealth is investing in tighter CI integration so environments become a standard part of the PR flow, automatic end-to-end suites on PR environments, and more dedicated test and load-test environments. They're also building toward tighter integration with an AI coding agent as a first-class part of their automated workflow — so an agent can spin up an environment, deploy a change into it, and validate that change on a real, prod-shaped stack before handing it back. As Glenn puts it, ephemeral environments are "the missing primitive that makes agentic development genuinely safe" at their scale.