What we learned building a platform NHS Derbyshire piloted
Children with neurodisabilities and long-term conditions can see 15–20 different health and care professionals across different settings — GPs, schools, therapists, specialists — each holding a piece of the picture, none holding all of it. That fragmentation is the problem EnrichMyCare set out to fix, and it's also the problem behind why neurodevelopmental referrals so often arrive incomplete, sit in a queue, and turn one appointment into three.
We partnered with Enrich Digital Technologies Ltd, founded by Saran Muthiah — a Specialist Children's Physiotherapist and NHS Clinical Entrepreneur — to build the software behind that fix. This post is about what we actually learned doing it, not a feature list.
The problem wasn't the app. It was the data arriving broken.
Our first instinct, like most engineering teams, was to think about this as a communication problem: give parents, clinicians, and schools one shared place to see a child's records, and the fragmentation goes away. That's true as far as it goes — and it's why the EnrichMyCare mobile app and web portal exist, built around a Care Team Hub, a secure records vault, and shared health, seizure, and sleep diaries.
But the deeper issue only showed up once we looked at the referral pathway itself. Referrals arriving from GPs, schools, and families were routinely incomplete — missing the specific structured information a clinician needed to triage accurately. That incompleteness, not the lack of a shared app, was what turned straightforward cases into multi-appointment ones.
Where AI actually earned its place
We didn't start with "let's add AI." We started with a narrower question: can a referral form catch its own gaps before it's submitted, instead of a clinician discovering them three weeks later in clinic? That's what the AI-assisted layer does — built-in checks that flag incomplete structured referral data from GPs, schools, and families before it ever reaches a queue.
The second piece was triage. Once referrals were arriving complete, we built clinician-supervised AI-driven triage logic to prioritise cases by complexity and risk — explicitly supervised, not autonomous. The AI narrows and ranks; a clinician still decides. That distinction mattered more to the NHS pilot teams than any accuracy metric we could report.
The lesson that stuck with us: AI-assisted triage is worth building when it removes administrative noise ahead of a clinical decision — not when it tries to replace the decision itself.
What the pilot showed
EnrichMyCare was piloted on a neurodevelopmental pathway with Derbyshire Healthcare NHS Foundation Trust. The results that mattered most weren't the ones we'd have guessed at the start of the project:
- Reduced by 40% once referrals arrived complete and pre-triaged, rather than being clarified back-and-forth after submission.
- Complex diagnoses that previously needed three or more appointments were, in a number of cases, resolved in a single visit — because the clinician had the full structured picture going in.
- The consumer mobile app was rebuilt on .NET MAUI mid-project for better cross-platform stability — a decision driven by real performance issues in production, not a technology preference.
What we'd tell another team starting this today
If you're building software for a care pathway — NHS-adjacent or not — the highest-leverage place to look first is the point where information changes hands: referral, handoff, discharge. That's usually where the actual delay lives, not in whichever screen looks the least modern. And if you're tempted to lead with AI as the headline feature, we'd push back gently: it's the completeness and structure of the data going in that made the triage logic worth having at all. Build that first.
We're proud of this one specifically because the number that matters isn't a usage metric — it's appointments a family didn't have to take time off work and school for, three separate times, to get an answer they could have had once.