This is a submission for the DEV Weekend Challenge: Community
The Community
AffineIA was built for the artisanal cheesemaking community — a diverse group that includes home enthusiasts learning to make their first fresh cheese, independent producers and affineurs looking for reliable technical references, and vocational trainers who need structured content to support their students.
This community has always transmitted knowledge through hands-on practice, oral tradition, and expensive professional training. Affordable digital tools tailored to their specific needs simply didn't exist. AffineIA was built to fill that gap — making deep cheesemaking knowledge accessible to anyone, at any level, at any time.
What I Built
AffineIA is an AI-powered assistant specialized in artisanal cheesemaking, built with Python, Gradio, and the Claude API (Anthropic).
It adapts to three user profiles:
- 🧀 Amateur — accessible recipes with practical guidance
- 🏭 Producer — precise technical sheets, including rennet types and process parameters
- 🎓 Trainer — structured recipes adapted for pedagogical use
Key features:
- 4 creativity levels for recipe generation: static reference (0), web-scraped (1), hybrid web+AI (2), and pure LLM generation (3)
- Expert chat — a free-form conversation interface with the cheesemaking assistant
- Recipe history — saved and retrievable across sessions
- Photo visualization — AI-generated image of the resulting cheese
- Knowledge base — a curated database of cheeses and their descriptions (currently being enriched with expert contributions)
- Notion integration — the knowledge base is synced with a Notion database for easy updates
- Secure access — multi-user authentication with hashed passwords
Demo
🔗 Live app (beta access): https://huggingface.co/spaces/volubyl/affineia
Access is currently by request during the beta phase — contact **u4gkuft9e@mozmail.com* to get your credentials or fill the form on [https://myriam-braconier.github.io/affineia-betatest]*
Code
The source code is currently private as AffineIA is in active commercial development.
The technical stack and architecture are detailed in the How I Built It section above.
How I Built It
Stack:
- Python — core application logic
- Gradio — web interface, multi-tab layout, custom HTML/CSS theming
- Claude API (Anthropic / Sonnet) — AI engine for recipe generation and expert dialogue
- Notion API — knowledge base management and synchronization
- Web scraping — dynamic recipe retrieval for creativity levels 1 and 2
- HuggingFace Spaces — deployment and hosting
- hashlib — password hashing for multi-user authentication
- GitHub — version control
The challenge I solved:
Cheesemaking is a craft that lives at the intersection of biology, tradition, and precise technique. Building an AI that gives reliable, contextually appropriate answers across three very different user profiles required careful prompt engineering and a domain-specific knowledge base — not just a generic LLM wrapper.
The 4 creativity levels were designed to let users choose between safety (reference recipes) and exploration (AI-generated originals), giving both beginners and experts a tool they can trust and enjoy.
AffineIA is currently in beta (before migration to stack Django/React. The knowledge base is being enriched through expert contributor feedback — if you know your cheeses, your input is welcome. 🧀
Top comments (1)
My app is bilingual French/English. Enjoy !