Some integration projects have flexible timelines. Some have commercial pressure. This one had a regulator. When a home-care and nursing services provider needed to synchronise appointment data with a government-mandated external platform, the timeline was not negotiable and the margin for error was zero. This is what a production-grade answer to that obligation looks like.
The mandate that changed everything
A home-care and nursing services provider operating at national scale supports a country-wide network of nurses, caregivers and the patients they serve at home. Its scheduling estate, built on Salesforce, handles appointment data for a patient population spread across an entire country.
A government mandate required that appointment data flow from that scheduling estate to a specific external platform, in near-real-time and at the provider's actual operating scale. This was not a discretionary improvement. It was a regulatory obligation with operational and compliance consequences if missed.
The integration backbone became the strategic answer to that obligation. The question was not whether to build it. The question was whether it would be built to a standard the regulator and the external platform team could both lean on.
Engineered for the scale that actually matters
Ampleshift joined in early 2024. The platform was engineered on MuleSoft CloudHub 2.0, picking up appointment changes from Salesforce, where both internal schedulers and patient-facing tooling feed in, and synchronising them in near-real-time with the mandated external destination.
The design target was 100,000 appointment events per week. Not a theoretical ceiling. The actual operating scale of a national home-care network. Every architectural decision, from error handling to observability to deployment patterns, was made to hold at that volume.
CI/CD pipelines on Azure DevOps were in place from sprint one, not bolted on at the end. Code quality gates and pipeline-enforced property encryption checks meant that missing values or unencrypted secrets were caught before they reached the version control repository. For a healthcare provider, a sensitive value committed unencrypted to version control is not a minor incident. It was removed as a class of risk from the engineering process entirely.
Engagement at a glance
- 100,000+ appointment events per week, validated end-to-end in UAT
- Zero-downtime deployment pattern via Azure DevOps and CloudHub 2.0
- Every category of failure caught and classified in UAT: connectivity, business and data errors
- Sensitive patient information masked from logs by default, from sprint one
- Approximately 2 years: early 2024 to February 2026
Observability that caught everything before production
Ampleshift deployed the full ELK stack alongside the integration platform from day one: Elasticsearch, Logstash, Kibana. Real-time monitoring and end-to-end traceability across every application in the estate, not as an afterthought but as a Day-1 capability.
During UAT, the observability stack caught every category of failure: connectivity errors, business errors and data errors. Each was classified correctly by the error-handling strategy. Technical errors that simply needed a retry were handled automatically. Business errors requiring intervention were surfaced to the team managing the data at its source, in Salesforce, rather than at the destination where they would have been harder to act on.
That distinction matters in a healthcare context. A missed appointment event is a missed appointment. The error-handling strategy was designed to protect against that outcome at every layer of the stack.
Healthcare data privacy, engineered in from the start
Sensitive patient information is masked from logs by default across every API in the estate. This was not a retrofit applied after the first audit question. It was a structural property of the logging strategy, applied uniformly from the first sprint.
For a home-care provider operating in a regulated environment with patient-level data flowing through the integration estate at scale, that posture was non-negotiable. Ampleshift treated it accordingly.
Senior expertise from day one. No billing games.
An Engagement Manager and Architect as the constants across the two-year programme. Up to three developers added at peak. Onshore, senior and elastic: scaled to the work rather than imposed at a fixed shape.
Common Assets were applied across every API: Template API, Common Flows, Parent POM, Global Error Handler. Every integration in the estate shares a single, consistent operational pattern. Reusable by the provider's own team. Operable without a dependency back on Ampleshift.
Validated end-to-end. Ready to go live.
End-to-end UAT validation was completed at the 100,000 events-per-week scale before hand-off in February 2026. The platform was production-grade and production-ready. Production go-live was held by a dependency on the external-platform team, a party outside Ampleshift's scope, not by anything in the integration estate itself.
That is a distinction worth stating clearly. Ampleshift delivered a platform the provider could switch on as soon as the external dependency cleared. No further engineering work required.
Two years. One mandate. A production-grade platform validated at national scale, with deployment discipline, observability and healthcare data privacy built in from sprint one. For home-care and nursing services providers, and any healthcare operator facing government-mandated data-sharing obligations, the integration backbone is not a discretionary improvement. It is the foundation that determines whether the regulatory obligation can be met at all.
