Taste Labs

Taste Labs is building the data and infrastructure layer to solve for subjective quality in artificial intelligence. The company’s stated mission is to eliminate “AI slop”—the generic, low-quality, and impersonal output generated by AI models—by creating a foundation that allows AI to understand and replicate taste, judgment, and style. Taste Labs recently emerged from stealth with an $18.5 million seed funding round co-led by Amplify Partners and CRV. Taste Labs' business model targets both AI developers and the applications built upon them. The company provides frontier AI labs with preference datasets, rubrics, and evaluation environments to assess and enhance their models' subjective output. Simultaneously, it offers software and tools to application-layer companies to help their AI agents produce more on-brand and creative results. The initial focus is on the domain of design. A core component of the company's strategy is its unique approach to data annotation. Instead of relying on general crowdsourcing, Taste Labs cultivates a high-quality network of vetted experts and “tastemakers” through a trust-based propagation model, where existing members nominate new ones. This curated community provides the high-quality human judgment necessary to train AI on subjective tasks, creating what the company calls an “infrastructure for operationalizing judgment.” Keywords: AI taste, data infrastructure, AI slop, subjective judgment, AI model evaluation, preference data, foundation models, generative AI, AI alignment, Thais Castello Branco, design AI, creative AI, brand alignment, annotation infrastructure, RLHF, aesthetic AI, AI quality, reasoning data, AI agents, enterprise AI