← Back to AIHealth Chat
Application Pack
Google Africa Applied AI Lab — AIHealth Chat
Application window: July 1 – August 31, 2026. Program: co-development with Google DeepMind at the Accra AI Community Centre (ACIC), Accra, Ghana, September–December 2026, with in-person participation required. Culminates in a demo day with startups pitching to Google AI Futures Fund and partner VCs.
This document is also the grounding source for the NotebookLM pitch deck. Every claim below is verified against the live product and repository unless marked [CONFIRM].
1. Verified Fact Sheet
Product
- AIHealth Chat — a 24/7 multilingual AI health assistant and automated health tracker, live on the web and on WhatsApp.
- Live web app: https://aihealthchat.qzz.io (also https://ai-health-chat.vercel.app). Installable PWA.
- WhatsApp channel: live via Meta WhatsApp Cloud API, with instant language switching (reply
EN, SW, FR, ES, or NL) and 6-digit linking of a WhatsApp number to the web dashboard.
- Positioning: evidence-based education and triage only — explicitly not diagnosis, prescription, or emergency response. Emergency-symptom messages short-circuit to localized “seek emergency care now” warnings.
Languages
- Full product localization in 5 languages: English, Kiswahili, French, Spanish, Dutch — across both the web UI and the WhatsApp bot.
AI architecture (already Google-AI-native)
- Primary web model: Google Gemini 2.5 Flash (with Gemini 2.5 Flash Lite handling triage classification and safety auditing).
- Three-stage safety pipeline on every message: triage classification (emergency / clinical symptom / general wellness / maternal & child) → response generation → post-response safety audit.
- Resilient fallback chains: web falls back Gemini → OpenRouter (Gemma) → Groq; WhatsApp runs Groq Llama 3.3 70B → OpenRouter; vision (photo triage of rashes, wounds, medication labels, lab results) runs on Mistral vision models.
- Stack: React 19 + Vite + TypeScript frontend; Vercel serverless backend with SSE streaming; Supabase for auth (email + Google OAuth), Postgres with row-level security, and storage; security headers (CSP, HSTS) and Supabase-backed rate limiting.
Features shipped and live
- Streaming multilingual health chat with voice input and text-to-speech read-aloud.
- Photo triage: upload an image of a rash, wound, medication label, or lab result for visual assessment.
- Automated vitals tracking: mentioning vitals in conversation (“BP today 120/80”) auto-extracts and saves structured records — blood pressure, glucose, weight, temperature, mood, symptoms, medication adherence, maternal/antenatal, immunization — with a trends dashboard (7/30/90-day and all-time) and per-source badges (web vs WhatsApp).
- EHR/SOAP-standard clinical visit summaries: chat history synthesized into chief complaint, history, reported vitals, symptom timeline, red flags, and advice given; shareable via expiring, revocable read-only links (
/report/[token]) for clinicians.
- Dark/light theme, PWA install, privacy/terms/about pages.
Traction and stage
- Stage: Launched — live public beta on web and WhatsApp.
- Monthly active users: early and real — currently in the 1–50 range.
- Recognition: 1st Place (USD 500), Power Learn Project (PLP) AI Hackathon, #PLPClassOf2024 — awarded for the project as “AI Health Assistant.” Public announcement: Facebook post
- Revenue: $0 ARR (free beta). Capital raised: $0 — fully bootstrapped, no investors.
Founder — Raymond Klanderman
- Solo technical founder; designed, built, and operates the entire product (GitHub: rayklanderman).
- Google Cloud Certified — Professional Cloud Architect (advanced-level certification issued by Google Cloud).
- Public-service digital systems experience, including work with the office of Hon. Esther Passaris (Nairobi Woman Representative), Kenya.
- Based in Nairobi, Kenya.
Roadmap (Now / Next / Later)
- Now: live multilingual web + WhatsApp beta; grow early users; harden reliability (tests, CI, security remediation backlog).
- Next: deepen WhatsApp reach (the dominant channel for the target population); medication reminders; clinician-facing portal.
- Later: USSD/SMS for feature phones, HL7 FHIR interoperability, offline/on-device inference, formal clinical validation partnerships.
Why Google AI, why now
- The product is already built on Gemini — deeper model access, credits, and DeepMind guidance directly compound what exists today rather than requiring a migration.
- Multilingual health guidance for African users is exactly where Gemini’s multilingual depth matters; Kiswahili support is live today.
- Mission fit: accessible, safe, localized health information for underserved populations — aligned with the Lab’s brief of AI products with distinct technological futures for Africa.
2. Draft Form Answers
| Field | Answer |
| Email | [YOUR EMAIL] |
| Company name | AIHealth Chat (operating as AIHealth Labs) |
| Company website | https://aihealthchat.qzz.io |
| HQ Location | Nairobi, Kenya |
| Field | Healthcare / Health Tech |
| One-sentence description | See below |
| Product demo / video | https://aihealthchat.qzz.io — strongly recommended: record a 2–3 minute video (Loom/YouTube unlisted) walking through sign-in → multilingual chat → vitals dashboard → SOAP summary → WhatsApp |
| Founder(s) | Raymond Klanderman |
| Social handles | [X / LinkedIn] |
| Product Development Stage | Launched (live public beta, web + WhatsApp) |
| ARR (USD) | $0 |
| Monthly Active Users | 1–1,000 |
| Total Capital Raised | $0–100,000 (bootstrapped: $0) |
| Investors | None — bootstrapped |
| Open to raising capital from Google? | Yes |
| Privacy & Terms | Accept |
One-sentence description (pick one):
- “AIHealth Chat is a live, multilingual AI health assistant on web and WhatsApp that gives underserved communities evidence-based symptom guidance in 5 languages (including Kiswahili), automatically tracks vitals from conversation, and turns chats into clinician-ready SOAP visit summaries.”
- Shorter: “A multilingual AI health assistant on web and WhatsApp — built on Gemini — that delivers safe, evidence-based triage guidance and turns patient conversations into clinician-ready visit summaries.”
3. NotebookLM Deck Prompt (final)
Upload as sources: this document, README.md, about.html (or its text), and screenshots’ descriptions if desired. Then run:
Act as a technology grant and accelerator program advisor preparing a founder
for the Google Africa Applied AI Lab (a Google DeepMind co-development program
in Accra, culminating in a demo day before the Google AI Futures Fund and top
VCs).
Using ONLY the uploaded sources, generate a 10-slide pitch deck with exactly
this structure:
1. Title — AIHealth Chat, tagline, and founder name (Raymond Klanderman)
2. The Problem — Access, cost/connectivity, and language gaps in health information for underserved African communities
3. The Solution — Live multilingual health assistant (Web PWA + 24/7 WhatsApp), evidence-based and safety-first
4. How It Works — Multi-tier AI engine, triage → generate → safety-audit pipeline, vision/image analysis, health tracking, and SOAP summaries
5. Why Google AI, Why Now — Built on Gemini; multilingual depth (English, Kiswahili, French, Spanish, Dutch); direct mission fit with the Lab
6. Traction — 1st Place PLP AI Innovation Hackathon win, live status on Web and WhatsApp, honest early user metrics
7. Roadmap — Near-term (Now), Mid-term (Next), Long-term (Later)
8. Founder — Technical background, Google Cloud Certified Professional Cloud Architect, public-service digital systems experience
9. The Ask — DeepMind model access, technical mentorship, open to Google AI Futures Fund capital
10. Closing — Live URL (https://aihealthchat.qzz.io), contact details, and call to action
For every slide provide:
- Slide Title: Bold, action-oriented statement (a concrete claim, not a generic section label)
- On-Screen Content: Maximum 3 bullets, each under 12 words, prioritizing concrete specifics from sources (e.g., Gemini, Groq, 5 languages, Supabase RLS, live URLs)
- Speaker Notes: 2–3 concise sentences for the founder to speak aloud
- Visual Suggestion: Specific screenshot, architecture diagram, or graphic to display
Rules: Do NOT invent numbers, partnerships, or claims absent from the sources.
If a slide requires a fact absent from the sources, write "[GAP: missing detail]"
instead of fabricating context.
Tone: Confident, concrete, and technically credible — concentrate technical
depth in slides 4–5; keep slides 2, 6, and 9 plain and human. Authentic to an
early-stage startup.
4. Eligibility notes to resolve before submitting
- The program looks for current full-time founders; if you hold another role, decide how to present your commitment.
- The co-development phase runs in person at ACIC in Accra, Ghana (Sept–Dec 2026) — confirm you can relocate/travel for it.
- Deadline: August 31, 2026 — about 3 weeks away.