Three clinician-facing builds,
recreated so you can watch them work.
Every delivered build on finecore.dev is clinician-facing and runs inside a hard privacy boundary: HIPAA for AleemIQ, on-prem-only data for CANN-DiS, home-network-only signals for ElderGuard. The real deployments hold data we cannot show, so this page recreates each UI with synthetic data. The workflows, guardrails, and pacing are what shipped.
Three recreated product walkthroughs. First, AleemIQ: an iOS anesthesia app fills in a synthetic case (laparoscopic cholecystectomy, ASA class III, a flagged difficult airway, a propofol TIVA plan), then surfaces three similar prior cases with citations to anonymous case records. Second, CANN-DiS: a clinician asks whether a patient on warfarin can start CBD oil; the chatbot streams an answer about CYP2C9 and CYP3A4 inhibition and INR monitoring, citing FDA DailyMed labels for cannabidiol and warfarin, while a status row notes inference is local and nothing leaves the network. Third, ElderGuard: a monitoring dashboard shows live respiration and heart-rate traces within the sensing ranges of 6 to 30 breaths and 40 to 120 beats per minute, room presence moving through the home, and an events timeline where a fall signature in the hallway pushes an alert to the caregiver app and resolves two minutes later as caregiver notified. All data shown is synthetic.
AleemIQ — anesthesia decision support
An iOS app for ambulatory anesthesia providers: anonymous case documentation plus precedent from prior cases, retrieved and cited at the point of care. Built for App Store submission.
The case shown is synthetic. In production, every record is documented anonymously and the precedent panel cites the matching case record.
CANN-DiS — cannabinoid drug-interaction chatbot
A clinician-facing chatbot for cannabinoid drug interactions. Local Gemma 2 inference with optional RAG over FDA DailyMed labels, FAISS search, and Whisper voice input. Answers are grounded in cited drug labels.
The exchange shown is synthetic. The deployment constraint was the product: where drug data could not leave the building, inference runs fully on premises.
ElderGuard — privacy-first elder monitoring
A cross-platform app paired with WiFi-CSI sensing hardware: contactless fall detection, vitals, and sleep analysis. No cameras, no wearables. Signals never leave the home network.
The feed shown is synthetic. iOS, Android, and HarmonyOS apps pair with the sensing hub; the fall alert path ends at the caregiver's phone.
Shipped, accepted, and still in production.
Three different products. One delivery system.
Data boundaries before feature work
Each build's boundary was set before feature work started: PHI handling for AleemIQ, on-prem inference for CANN-DiS, home-network-only signals for ElderGuard. BAAs and audit logs where required.
Grounded in cited sources
AleemIQ cites prior case records; CANN-DiS cites FDA DailyMed labels. Every answer carries its source, and an eval set proves citation accuracy before rollout.
Clinician-in-the-loop
The tools document, retrieve, and surface. The clinician decides. Outputs that touch care routes through clinician review, and the audit trail records who approved what.
Your build is next.
The same Acceptance-Led Delivery system that shipped these three is what a new engagement gets: one acceptance metric in your language, a weekly demo, and an eval set you keep.