This is a submission for the Gemma 4 Challenge: Build with Gemma 4
What I Built
A customer walks up to a welder in Nairobi with a Pinterest screenshot. "I want this gate. How much?"
The welder squints at the phone. Does mental math. Guesses a number. Scribbles it on a scrap of cardboard. No material breakdown. No line items. The customer walks away and finds someone who looks more professional.
This happens thousands of times a day across Kenya's jua kali sector. "Jua kali" translates to "hot sun." It's what we call the 15 million informal artisans who build gates, furniture, window frames, and cabinets outdoors. They're skilled fabricators who lose work because they can't produce a quote on the spot.
Jua Kali Quote fixes this. Photograph whatever the customer shows you: a sketch on paper, a screenshot from Instagram, a photo of their neighbor's gate. Gemma 4 looks at the image, figures out what needs to be built, and generates a full quotation with materials, quantities, and pricing in Kenyan Shillings.
The fundi taps any line item to adjust a price they know better. The totals recalculate live. They hit "Share on WhatsApp," and the customer has a professional quote in their chat. The whole interaction takes under a minute.
Every quote is saved locally, so the fundi builds a history of past jobs they can reference for repeat customers or similar work.
A confidence indicator tells them upfront: "materials ±15%, labor ±20%." Honest about what's an estimate and what's exact.
It handles welding, carpentry, masonry, plumbing, electrical, and painting jobs.
Demo
Code
Jua Kali Quote
AI-powered quotation generator for Kenyan artisans (fundis). Upload a photo of a sketch, Pinterest screenshot, or reference image and get a professional quotation with materials, labor, and pricing in KES.
Built with Gemma 4 26B A4B (MoE) via Google AI Studio.
Setup
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
Copy .env.example to .env and add your Google AI Studio API key:
cp .env.example .env
Get a free API key at https://aistudio.google.com/apikey
Run
source venv/bin/activate
uvicorn main:app --host 0.0.0.0 --port 8000
How it works
- Upload a photo (sketch, Pinterest screenshot, catalog image, existing item)
- Select the trade (welding, carpentry, masonry, etc.)
- Gemma 4 analyzes the image and generates a structured quotation
- Edit any line item price if needed
- Share via WhatsApp or print as PDF
Stack
- Python + FastAPI
- Gemma 4 26B A4B IT (Google AI Studio API)
- Vanilla HTML/CSS/JS (mobile-first)
- localStorage for…
How I Used Gemma 4
I went with Gemma 4 26B A4B, the Mixture-of-Experts variant. Three reasons.
Vision that understands intent, not just pixels. The model doesn't just see "a rectangle with vertical lines." It recognizes that's a gate design with panels, estimates it at roughly 12 feet wide, and knows that means four hinges, not two. It reads hand-drawn sketches and blurry phone screenshots equally well.
Reasoning that produces consistent numbers. This isn't image captioning. The model estimates material quantities, looks up realistic Kenyan prices (square tubes at KES 3,500, electrodes at KES 1,500, fundi day rate at KES 1,200), then adds everything up correctly. The grand total actually matches the sum of its parts.
MoE efficiency for real-world speed. 26B total parameters, but only 3.8B active per token. A fundi standing in front of a customer doesn't want to wait 30 seconds. The MoE architecture gives near-31B quality while keeping response times practical on a mobile connection.
The technical setup is minimal:
Photo upload → FastAPI backend → Gemma 4 (Google AI Studio API)
→ Structured JSON response → Rendered quotation → PDF or WhatsApp
The system prompt tells Gemma 4 to act as a Kenyan construction estimator with current market prices. No RAG, no database of prices, no fine-tuning. The base model already knows enough about Kenyan hardware stores to produce quotations that a real fundi would look at and say "yeah, that's about right."