Workshop 01 · Discovery & alignment
A shared record of the vision, product concepts, and business context discussed in the first workshop — used to confirm alignment before Workshop 2 is scheduled.
What we heard. SwissDnaCode wants to turn today's isolated ChatGPT chatbot into its own medically and legally secured platform — a team product with defined roles for the founders, staff, and clients, grounded in functional / regulative medicine and real medical oversight.
The MVP. The first release is focused on the chatbot: a safely-guided client bot, with roles, graduated permissions, and credit-based usage. The team-facing doctor bot, the native app, and everyday scans are part of the vision but not in the MVP.
What we need from you. Review this summary and confirm alignment, set a target go-live date, and decide whether the native app is in the MVP or a later phase.
Today there is a single, isolated chatbot built on ChatGPT, used only by Dr. Farkas with her clients. It is not connected to any other system and offers little individualization. The goal is to turn it into SwissDnaCode's own legally and medically secured platform — a team product with defined roles for administrators, staff, and clients.
Dr. Nadine Farkas
Medical lead
Owns all medical matters and responsibility. Uses the bot clinically with clients.
Dante Scherrer
Tech & ops lead
Tech, support, website, communication, advertising. Requirements approved jointly.
SwissDnaCode comes from functional / regulative medicine. Clients are deeply interested in their own genetics and trust medically-supervised guidance — and they tend to share more openly with the bot than face to face. The long-term ambition is for Dr. Farkas to step back from day-to-day practice and focus the company on DNA and blood-cleansing expertise, with a later option to license the platform to other therapists.
Clients since Feb
10
On the current chatbot
Accompaniment
3 mo
Then optional self-serve
Usage limit
Credits
Metered per use
Feedback survey
5★ · 5 Q
Proposed measure
10 clients since February. They use the current ChatGPT-based chatbot, value the support, and tend to share more openly with the bot than in person.
Accompaniment — 3 months. SwissDnaCode accompanies the client at the start (preferred, if desired). After three months the client can continue unaccompanied, using only the chatbot. This is an optional transition point, not a fixed term.
Usage limit — credits. Client usage of the system is metered through a credit system.
Feedback survey — 5★ · 5 Q. A proposed success measure: a five-star rating with about five questions. It is a feedback tool rather than a confirmed KPI — see the "measurable KPIs to define" note below.
A core part of the vision is everyday use from home — a main requirement — through a simple mobile experience that signals the real medical expertise behind it. Whether this native app ships in the MVP or a later phase is still to be decided (see roadmap). Typical everyday questions it would answer:
Photograph a restaurant menu and ask which dish fits their DNA.
Scan a product while shopping in the supermarket.
Check whether a particular sport suits them.
Ask general life and wellbeing questions.
Future: useful answers could be saved — a scanned menu becoming a permanent menu card, a fitness plan getting its own "Fitness" navigation point. Categories such as fitness and nutrition are predefined, and questions are auto-categorized (stress, sleep, nutrition…).
Questions are sorted into three categories, each with its own restriction level (general, detailed, or blocked). The agreed starting point is to block everything and open capabilities gradually — so the levels below are a proposed direction, not yet a fixed mapping.
Lifestyle
Supplements, nutrition, sleep, stress, fitness
Proposed: generalMedical
Findings, SNP correlations, diagnoses
Proposed: limited / detailedMedication
Therapy, antibiotics, dosing
Proposed: blocked for clientsHard guardrail: the client-facing AI must never provide medical therapy. It always shows "I am not a doctor / ask Dr. Farkas". Example — a client with bronchitis is referred to a specialist and cannot ask which antibiotics to take.
A key idea raised in discovery is splitting the assistant in two, so the medical depth lives with the team while clients stay safely guided. The client bot is the MVP focus; the doctor bot originated as a proposed solution and is planned for a later phase.
Client bot MVP
Doctor bot (team) Future
A chat summary can be shared with the attending therapist only with the client's consent — a data-protection-sensitive point, as clients should not feel monitored.
A concrete example of the in-practice workflow, once a patient has completed the SwissDnaCode code:
Prioritize the genetics. The bot summarizes which tested DNA SNPs are most important for this patient.
Correlate to diagnoses. It links those priorities to the diagnosis list the patient brings in.
Check the medication list. The patient's medications are reviewed against the relevant SNPs.
Answer patient questions. For example, the recommended food list during gut cleansing.
Attach findings. Depending on the case, gut and lab findings are uploaded to the record.
TODAY
Isolated bot
ChatGPT-based, Farkas only, not integrated.
MVP · FOCUS
The chatbot
Client-facing bot with roles, permissions & credits.
AFTER 3 MONTHS
Self-serve
Client renews unaccompanied, chatbot only.
FUTURE
Scale + vision
Doctor bot, everyday scans, affiliate, Zoom, certification.
The MVP is focused on the client-facing chatbot — the conversational experience and its medical safeguards ship first. The team-facing doctor bot is a later phase, and whether the native mobile app (and everyday scans for menu, product, sport) is part of the MVP or a later phase is still to be decided.
New features are reported to SwissDnaCode for marketing; clients are told via in-app notifications, and marketing partners are briefed early before each go-live.
Differentiators (USP)
Feature requests
These are SwissDnaCode's intended differentiators. They have not yet been fully tested against the market — the competitive landscape is still an open subject.
Early finding — closest competitor: SelfDecode (DecodyGPT) selfdecode.com ↗
SelfDecode already offers an AI health assistant that combines a user's DNA, blood results, and symptoms to answer everyday health questions with personalised recommendations — and it shows a "not a substitute for a doctor" disclaimer. That is almost exactly SwissDnaCode's client-bot concept, already on the market.
Implication: an AI chatbot built on your DNA is not by itself a differentiator. SwissDnaCode's defensible ground is the combination most rivals lack — a doctor-led model (vs. coach- or therapist-led), Swiss data residency, and supervised onboarding. The "real doctors" USP is the one worth leaning into and validating first.
Products
Business model
Data-protection and legal documents are provided by Mendelio as drafts; final responsibility remains with SwissDnaCode.
Items that could significantly affect scope, timeline, or cost, and should be assessed early.
Medical-device certification. If the software requires Swissmedic / EU MDR certification, it becomes a major driver of timeline, cost, and architecture. An early legal/medical assessment should confirm whether it applies to the MVP.
Doctor bot liability & data protection. The proposed doctor bot would hold full patient data and answer medication questions — a high-liability capability. Before it is built (a later phase), it needs a clear GDPR / Swiss FADP lawful basis, consent flow, access controls, and audit trail.
Swiss data residency vs. LLM hosting. "Servers in Switzerland" is a headline USP, but a chatbot typically calls a EU-hosted model. If patient data leaves Switzerland to reach the model, the promise and the architecture conflict — an early decision is needed.
Key-person dependency. Currently, medical knowledge and clinical workflow are concentrated in Dr. Farkas. The platform should capture and scale that knowledge rather than depend on her availability.
Scope creep. The "future" bucket (native app, scans, affiliate, Zoom, Health Academy) is large and tends to pull into the MVP. The chatbot focus should be protected through the MoSCoW exercise.
Health-claims & terminology. Terms such as "blood cleansing" and related health claims need legal/medical review before appearing in client-facing or marketing material.
This summary captures the discovery discussion. The stakeholder map and annotated prototype review follow as separate artifacts, and a few items remain open before Workshop 2.
The next workshop goes below the surface: who the users really are, the problems they face, the features that solve them, and the boundaries of the first version. Below are the questions we'll work through together, tailored to SwissDnaCode.
Meeting 03 · Users & problems
Meeting 04 · Features & scope
Workshop 2 deliverables