Generative modeling for recipe generation: Success stories
summary
Thoughtworks applies generative models to R&D for consumer goods, claiming faster, repeatable product discovery with higher success rates and less wasted raw material. The team builds a production pipeline using a predictor+generator tandem that trains on consumer panel feedback in an active loop, leverages active learning to work in low data regimes, and runs as a containerized Python stack with a centralized schemaed database and event-driven orchestration. Deployments across snacks, beverages and whisky projects report practical gains.
AI-written summary · about
chapters (9)
- 0:00Generative AI in food R&D
- 1:07Generative models overview
- 7:55Product developers' role
- 13:42Testing generative AI with large language models
- 17:17Traditional recipe optimization
- 21:11Recipe experimentation challenges
- 23:17Data ecosystem challenges
- 25:24AI adoption in R&D
- 27:56Success stories and applications
related talks
17mGenerative AI: Keeping it real
22mGenerative AI - A Pragmatic Business View
29mGenerative AI: Path to Value in practice
45mTrusting AI to Modernize Software at Scale
27mGenerative AI for technology leaders
15mPractical Evolution of Data Science with GenAI
01h 20mGenerative AI for Businesses
22mGenerative AI & its Transformation of Business, Research, and Society