3 articles
An AI cooking assistant where safety is paramount. Exploring deterministic safety layers, vector search, and when NOT to use LLMs.
Building responsive voice-guided cooking experiences
Deep dive into our runtime LLM architecture, speech synthesis pipeline, and observability systems that power conversational cooking guidance.
How we process 35,000+ recipes with 99.9% accuracy
Deep dive into our 7-stage recipe processing pipeline with intelligent routing, cost optimization, and comprehensive safety mechanisms.
Why we don't trust LLMs with allergies
Building deterministic safety layers with hybrid NLP, streaming architecture, and rule-based allergen detection for health-critical AI applications.