Healthcare
Healthcare software is judged in the few seconds a clinician has between patients. A system that is correct but slow gets worked around, and a workaround in a clinical setting is a safety problem rather than an adoption metric.
Integration is the hard part
Teams new to the domain usually budget for building the product and are surprised by the interfaces. A patient record is scattered across an EHR, a lab system, an imaging archive, a scheduling tool and a billing platform, connected by HL7 v2 messages, FHIR APIs of varying completeness, and occasionally a nightly file drop nobody will admit to owning.
None of it is conceptually difficult. It is simply where the schedule goes. We would rather map your actual interfaces during discovery — including which fields are reliably populated in practice, as opposed to which are theoretically supported — than discover in integration testing that a required identifier is optional in your instance.
Where we work
EHR and clinical system integration
HL7 v2 and FHIR interfaces built for the messiness of real feeds, including partial messages and fields that are optional in practice.
Patient-facing platforms
Portals, intake and telehealth designed for people who are unwell, distracted or using a borrowed device.
Clinical workflow tools
Software used between patients, where a saved click is worth more than a redesigned dashboard.
Claims and revenue cycle
Coding, eligibility and claims pipelines where a rejection has to be explainable and correctable, not just logged.
Analytics on clinical data
Cohorting, quality measures and reporting built so every figure traces back to the encounter behind it.
Medical device and remote monitoring
Ingesting device telemetry with the gap-handling and ordering guarantees that intermittent connectivity forces on you.
Constraints that shape the work
Interfaces you do not control
HL7 v2 feeds with site-specific quirks, FHIR endpoints at different maturity levels, and vendors whose release schedule is not yours.
Downtime is clinical
A maintenance window that inconveniences an office is a patient safety event in a ward. Migrations are designed around this, not scheduled around it.
Audit as a first-class feature
Who saw which record, when, and why. Retrofitting this after go-live means reworking every data path that touches identifiable information.
Consent is structural
Purpose limitation and revocation have to be modelled in the schema. A consent flag checked at the UI is a policy, not a control.
Privacy is an architecture decision
HIPAA obligations are easier to meet when the system is arranged so that most components never touch protected health information. Segmenting identifiers from clinical detail, keeping PHI out of logs and analytics events by construction, and making the audit trail a first-class table are all cheaper decided early than retrofitted.
We engineer to HIPAA requirements and support your audits and assessments. We are not an auditor and do not certify anything — where you need a formal attestation, that comes from a qualified assessor, and we make sure the evidence they will ask for already exists.
- PHI minimised by design, and kept out of logs and telemetry
- Access controls tied to clinical role, with break-glass recorded
- Immutable audit trail of who saw which record, and when
- De-identification for analytics that survives a re-identification review
- Business associate obligations understood before the build, not after
When not to hire us
We hold no healthcare certifications and we are not a HIPAA-covered entity. Where your programme needs a supplier with an accredited compliance posture, we are not that supplier, and we will say so before you spend time on us.
The pattern we most often decline: an integration project scoped without the clinical staff who will use the result. Software that is technically correct and clinically unusable gets routed around within a month, and no amount of engineering rescues it.
Common questions
Yes, and it is now standard practice, provided the business associate agreement is in place and the configuration matches it. The engineering questions have been settled for years; the work is in the evidence and the discipline of not letting PHI leak into places nobody classified — logs, analytics, support tooling.
Related work
Bring us the part that is stuck.
Send the constraint you are working around and we will come back with an approach, a rough shape of the work, and what it would take to hand it over.