Club de Lausanne
AI Applied Advisory · Switzerland
For Patrick · mylife Diabetes Care
Pilot landing for mylifeThree lenses · one Pilot desk · exit pack your team keeps
A short pilot to make Product Care faster — without touching therapy AI
Start with the myLoop portfolio, then how mylife competes on AI — plus an Embedded Software Engineer case. Glossary in Appendix. Same Club de Lausanne framework throughout.
Where a pilot would sit on your portfolio
Choose a part of myLoop — the AID system, the pump, sensor partners, Product Care, or the independence change.
Below is a preview of the Pilot desk: the shared space a short mandate would install and fine-tune with your teams.
What this is
A working desk for AI-assisted Product Care — not a patient app, not a firmware IDE.
It shows how requests become clearer inputs, safer AI use, and decisions your experts still own.
Who uses it inside
Sponsors like Product Management first; then the people they pull in for a slice — R&D, QA, RA, Supply, Marketing, L3.
Specialists stay specialists. The desk connects handoffs; it does not replace clinical or QMS judgment.
Process & livrable
Process: Intake → chain map → harden inputs → weekly cadence → close.
Livrable: an exit pack the team keeps — map, templates, AI rules, cadence card, and a steering deck (e.g. Gamma).
A desk tuned to one portfolio slice (you choose which)
Chain map + handoff language Product can reuse with R&D, QA, RA, Supply, Marketing
AI rules under your governance (public / private / gated)
Cadence card + exit pack — not open-ended hours
Availability: beginning of September, or earlier if useful
Scope note
In scope: Product Care packs, PMS / CAPA assist, launch readiness, cross-functional handoffs, team enablement.
Out of scope: firmware ownership, QMS / release sign-off, clinical decisions — those stay with mylife specialists.
Walk the product switcher in 20 minutes — then lock one slice for the pilot demo tenant.
Outline copied — paste into Gamma (Create from text).
Where mylife stands on AI — and where a pilot buys speed
Competitors fight on therapy AI (closed-loop algorithms).
Most also race quietly on enterprise AI — how Product, Quality, and partners move decisions.
Advisory does not rewrite CamAPS. It closes the second race so mylife stays fast while the portfolio stays complex.
Public-signal benchmark (sites, partnerships, market guides 2025–2026) — not insider diligence.
Scores are relative positions for a Product Care conversation, not clinical superiority claims.
Layer A · Therapy AI
AID / closed-loop in the product
Adaptive dosing, smartphone control, CGM choice.
mylife myLoop + CamAPS FX is already in the peer set with Medtronic SmartGuard, Tandem Control-IQ, Omnipod SmartAdjust, iLet, twiist/Tidepool.
This layer is owned by R&D and algorithm partners — not by advisory.
Layer B · Enterprise AI
How the company uses AI to run Product
Handoffs, PMS/CAPA packs, partner impact sheets, training content, launch readiness after independence / rebrand / Glooko data shift.
This is where mid-size AID players often lag megavendors — and where a Pilot desk creates measurable advantage.
Peer set (AID portfolio)
Who patients and HCPs compare you to — and what public signals say about AI on each layer.
Player
AID / therapy AI signal
Digital / data signal
Therapy AI
Enterprise AI*
mylife Diabetes CaremyLoop · YpsoPump · CamAPS FX
Smartphone AID; adaptive CamAPS; multi-CGM (G7, Libre 3 / 3 Plus); pregnancy & young ages in claims set
Glooko as primary clinic data layer; legacy mylife Cloud / Software / Uploader phase-out by end-2026; independence + rebrand in flight
Lead / peer
Behind*
MiniMed (ex-Medtronic)780G · SmartGuard
Pump-integrated SmartGuard; Meal Detection™; large installed base; ecosystem CGM path
Scale digital org; CareLink heritage; parent Medtronic AI / digital muscle even post-split
Lead
Ahead
Tandemt:slim X2 / Mobi · Control-IQ+
Predictive Control-IQ; mobile bolus; multi-CGM path expanding
Tubeless AID leader; multi-CGM rollout; T2 expansion story
Strong consumer digital brand; Glooko / clinic integrations at scale
Lead
Ahead
Beta BionicsiLet
Minimal meal input (no carb counting) — different AI UX bet
Smaller org; focused clinical narrative
Peer
Peer
Sequel · twiistTidepool Loop / Diabeloop path
New entrant (2025–26 awards); open-algorithm story; Apple Watch angle
Startup digital speed; still proving scale ops
Rising
Peer
Therapy AI (product)
On the board — not behind
myLoop’s adaptive, smartphone-based CamAPS stack with multi-CGM choice is a credible differentiator vs closed ecosystems.
Gap risk is commercial share and tubeless roadmap timing — not “no AI in therapy.”
Enterprise AI (how you run)
Likely behind megavendors*
Independence, legal manufacturer, rebrand, multi-partner CGM, and Glooko migration multiply handoffs.
Big players absorb that with larger digital / PMO benches. That is the lag a Product Care pilot can shrink in 30–90 days.
How Club de Lausanne makes mylife more competitive
AI Applied Advisory
Protect therapy AI focus.
Keep R&D and CamDiab on algorithm / pump excellence. Advisory owns the operating system around Product Care — not the dosing loop.
Win the complexity race.
Multi-CGM + partners + country rollout is a mylife strength on paper. Enterprise AI (Pilot desk) turns that into faster impact packs than rivals who drown in their own scale.
Absorb independence without losing speed.
Legal manufacturer / naming / labeling / web / RA continuity is a temporary tax. Cadence + exit packs prevent the tax from becoming permanent lag vs Tandem / Insulet / MiniMed.
Make Glooko migration a Product Care asset.
Clinic data consolidation is table stakes. The competitive edge is clean internal handoffs and training so HCP-facing teams do not stall during cutover.
Buy expertise outcomes, not AI theatre.
Sponsors get maps, AI may/must-not, cadence, and decks teams keep — while megavendors burn cycles on generic copilots without QMS-aware framing.
What this page is not
Not a clinical head-to-head on time-in-range. Not a claim that consulting replaces algorithm R&D.
It is a Product Management lens: where public competition sits on AI, and where a Club de Lausanne Pilot desk buys relative speed on Layer B.
I read the Embedded Software Engineer posting as a map of how your teams already work —
SCRUM, CI/CD, IEC 62304, AI acceleration.
Below is the workbench a pilot would fine-tune with you: your terms on the left, advisory plug-in on the right.
Pilot fit using delivery vocabulary from your postingCanvas 4
Worked through it
Mapped posting language to pilot plug-ins
Framed BA · PMO · safe AI enablement
Drafted Context ↔ Plug-in view for Product
Summary
A pilot installs a safe AI workbench around how you already deliver —
SCRUM, CI/CD, IEC 62304, AI acceleration — so experts stay experts and inputs get clearer, faster.
Context
Embedded C / C++STM32RTOSBluetooth ATT/GATTIEC 62304AI accelerationTDDSCRUMCI/CDGit
Plug-in
Business AnalysisPMO / AI×PMFramework designDashboard fine-tuneEmployee trainingAPI connectorsCybersecurity agentOutput enablement
ContextPlug-in
IEC 62304
Docs, traceability, test assist under your QMSProcess acceleration for the team
SCRUM · CI/CD · TDD
Cadence + clearer handoffs; AI where it speeds reviewsEngineers keep technical judgment
Owned by your embedded expertsI connect the chain and enable supporting outputs
Cross-functional delivery
BA + PMO frame across the value chainExpertise outcomes, not substitute engineering hours
/pilot-fit ×
Business Analysis
intake
PMO / AI×PM
cadence
API connectors
systems
Cybersecurity agent
risk
Output training
docs · decks
Service offer (pilot-shaped)
Show a safe AI workbench, then fine-tune it on your premises
Business Analysis first — then train teams to generate documents, rules, presentations
Connect systems by API (public one-way · private protected)
Cybersecurity topic agent (public security knowledge + your policies)
Availability: beginning of September, or earlier if useful
Scope note
This page positions an advisory pilot around your delivery vocabulary. Embedded firmware ownership stays with your specialists. I install the frame that makes their expertise travel.
Happy to walk this page in 20 minutes — and adjust the demo tenant to language your Product team already uses.