Awaiting customer approval. This page isn't marketed anywhere.
Chris Irving
Engineering Manager
Jane is an all-in-one practice management platform for health and wellness clinics. It helps clinics handle everything—booking, charting, scheduling, billing, and getting paid—securely and online.
Manual, error-prone delivery process (Excel "merge queue") limiting deployments to ~3 per day.
Scaling from 100 to 250 developers made serial deploys too slow; CI became the bottleneck.
Flaky tests and lack of observability blocked confidence in parallelism and batching.
When Chris Irving joined Jane, the deployment process was almost entirely manual. They tracked pull requests and releases using spreadsheets and ad hoc communication. There was no unified merge queue, no observability into what was going live, and no clear way to determine whether a single PR or a batch of 10 was causing issues.
This manual model initially worked, but as the developer headcount approached 100, it quickly became unsustainable. Bottlenecks surfaced everywhere: code reviews stalled waiting for someone to manually decide merge order, unexpected conflicts delayed releases, and developers had little confidence in what was actually deployed.
Recognizing these constraints, the team moved to a more structured approach. They began with serial CI deployments through Mergify. This shift provided them with a firm baseline: a predictable deployment cadence (up to 20–22 deployments per day within a 12-hour window), clear visibility into pipeline stages, and fewer merge conflicts. But most importantly, it laid the foundation for future scaling.
Our merge queue was basically a spreadsheet. We were lucky to get three deploys out a day, and we had no insights into what was shipping. With Mergify, we transitioned from manual chaos to over 20 safe deploys per day. Now we're pushing into parallelism, and the bottleneck has shifted to CI — exactly where it should.
Chris Irving
Engineering Manager
With the serial approach stabilized, Jane's next challenge was to increase throughput further. But moving to parallel and batched merges isn't trivial — you need robust guardrails to avoid breaking the main branch. Jane's team approached this with caution:
They introduced batching, grouping multiple PRs to execute tests once across the group rather than per PR. This reduced redundant CI runs and conserved resources.
They built guardrails and validation layers so that PRs entering a batch would still be individually validated (e.g., performing quick checks first, followed by deeper tests).
They incrementally expanded the parallelism: starting with safe modules, small batches, and gradually scaling to more aggressive workflows.
Today, Jane is phasing out pure serial deploys in many modules. They're increasing merge throughput, reducing queue wait times for developers, and reclaiming engineering cycles previously lost to CI churn.
We’re doubling down. Mergify isn’t just a tool anymore — it’s becoming a critical part of our delivery function. As we grow, reliability and observability are non-negotiable.
Chris Irving
Engineering Manager
Even with batching and parallelism in place, Jane recognizes that no pipeline scales if flaky tests and blind spots persist. Their future roadmap is ambitious:
Flaky test detection & quarantine: identifying tests that fail intermittently, quarantining them so they don’t block merges, and triggering remediation (retry or fix).
Historical attribution: tracking when a flaky test first appeared, which PR introduced it, and who owns it — so remediation can be prioritized.
Preventing new flakies: integrating test stability metrics into the merge queue so only high-confidence tests run as blocking checks.
Deep observability: providing engineers with full job-level, test-level, cluster-level metrics — understanding where time is spent, which tests are slow, and when regressions begin.
Cultural integration: making test hygiene part of the dev workflow, not separate. Engineers will get early feedback on test flakiness, backed by signals from the system.
With these efforts in flight, Jane is carving a path toward reliably merging 250+ PRs per day, gradually decomposing monolithic workloads, and safeguarding their main branch even under heavy scale.
