From the health profile a patient builds on their own phone to the coded brief a clinician reads before the consultation, here is how Doogie works, and the safety, evidence and data protection behind it.
The problem
Take South Africa as one example among many: a person's health story is spread across a public clinic, a private GP, a pharmacy and a hospital that never shared a record, and most of it was never written down at all. In countries with mature health records it sits inside systems built for episodes rather than prevention. Either way the consultation starts cold, and the family-health and lifestyle context that makes prevention possible is the first thing the clock takes away.
Sources: Statistics South Africa, General Household Survey 2024; WHO (2023); Irving et al., BMJ Open (2017). Figures shown for South Africa as a worked example; the same fragmentation appears in most health systems.
How Doogie works
Doogie handles the part that eats the clock, and it does the job thoroughly every time, so the clinician arrives already informed and can spend the consultation on judgement and care.
Doogie talks with the patient in plain language and adapts each question to the last answer, so it never feels like a fixed form. It covers the current concern, medicines, lifestyle and the family-health history that usually never reaches a record.
Feels like: a calm conversation, not a questionnaire. On a basic smartphone, on a slow connection, in the patient's own language.
Behind the conversation, Doogie organises what the patient reported against recognised clinical guidance and marks anything that warrants a clinician's attention. That work is there to inform the clinician, never to reach a conclusion of its own.
Grounded in: established clinical knowledge summaries and primary-care pathways.
The practice receives a concise, structured brief: what the patient reported, what has changed since last time, and what to look at first. It arrives coded in SNOMED CT and ready to drop into the record.
Every line traceable: each item links back to what the patient actually said.
Built for the market
Doogie is designed for the hardest conditions first: intermittent connectivity, low-end phones, many languages and records that never followed the patient. That is a design brief, not a launch region. A product that works there works anywhere.
Built for an entry-level Android handset held in one hand, not a desktop in a consulting room. No app store download required, and no new hardware for the practice.
Any smartphoneText-first, small payloads, and a conversation that survives a dropped connection and picks up where it stopped. Data cost is treated as a clinical access issue, because it is one.
Works on a weak signalPatients complete their profile where they already talk to everyone else, so the practice is not asking them to learn a new channel before they have even arrived.
Meet patients where they areA patient explains their health best in their own language. Doogie converses in the patient's language and hands the practice a brief in the language the clinician works in.
Patient's own wordsMost people move between a public clinic and private care over a lifetime. The health profile follows the patient across both, so context is not lost every time the setting changes.
Mixed systemsThe patient is told plainly what is collected, who will see it and why, in language they can act on. Consent is granted per practice and can be withdrawn, because here trust is the whole sale.
Consent by designSafety, the spine of everything
Doogie never closes a clinical encounter on its own. Every brief is a draft for a qualified clinician to review, accept, edit or discard. Accountability stays with the human.
Doogie’s replies pass through rule-based code that runs before a patient sees anything, does not consult a model, and does not vary between runs. Today it enforces one thing absolutely: Doogie will not comment on medication or dose. A deterministic red-flag screen is the next control we are building, and until it is measured we will not claim it.
Doogie stays inside collecting and structuring what the patient reports. It will decline to advise on treatment or medication, and says clearly when something is outside what it can safely help with.
Escalation is designed to stop the conversation. When Doogie recognises something urgent, it ends the history, shows the local emergency number and urgent-care options, and alerts the practice. It never asks a patient in danger to keep answering questions. Recognition today is model-mediated and has not yet been independently measured, so we describe how it is built rather than how well it performs.
Clinical evidence & quality
Every release is evaluated against a defined quality framework, benchmarked with practising clinicians in the markets we serve. These are the dimensions we grade, and the bar we build toward.
We evaluate Doogie the way medicine evaluates itself: against expert consensus and documented pathways, with safety weighted above everything else.
Illustrative targets from our evaluation framework. Formal clinical benchmarking with partner practices is planned; we will publish results.
Trust & data protection
Doogie speaks the standards health systems already run on, and treats data protection and clinical safety as engineering requirements, not afterthoughts. POPIA and GDPR from the same architecture, and adaptable to the regime wherever a practice operates.
Built to the strictest of the data-protection regimes we operate under, POPIA and GDPR, from the ground up: lawful basis, purpose limitation, data minimisation and full auditability, with patient data held in infrastructure in the patient's region and encrypted in transit and at rest. The same architecture meets GDPR for UK deployments.
POPIA · GDPR · Data residencyThe health profile belongs to the patient, not to a practice or a funder. Consent is granted per practice, in plain language, and can be withdrawn. Every access is logged and visible.
Consent by designFindings coded against SNOMED CT via the SNOMED CT terminology services, so the record holds clean structured data instead of free text to retype.
SNOMED CT terminology servicesStructured output as FHIR R4 resources, so a Doogie brief slots into modern clinical systems without bespoke plumbing.
Standards-firstA clear path to structured write-back into the practice management and health record systems clinics already run, so briefs land where clinicians already work.
FHIR write-backDeveloped under a formal clinical risk-management process aligned to recognised clinical safety standards, with a clinical safety officer and a living hazard log.
Clinical risk management