Rubric AI

Post-training research and product lab

Winter 2026ActiveB2BArtificial IntelligenceData LabelingBig DataInfrastructureAISan Francisco, CA, USA

Company

Rubric AI is a post-training research lab (backed by Y Combinator) building the human and computational layer that turns expert judgment into training signals for frontier models and agents. We believe that shaping the world's most consequential AI systems requires both purpose-built training environments and exceptional judgment.

Verdict

High Signal

Market Opportunity

Post-training data and RLHF infrastructure for frontier AI labs is a massive and rapidly growing market — Scale AI alone was valued at $13.8B. ICP is clear: frontier AI labs, regulated enterprise verticals (healthcare, legal, finance). B2B with clear monetization via data contracts and platform licensing.

Medium Signal

Founder Signal

Pragya Saboo is a non-technical solo founder — a significant concern for a deep-tech infrastructure play. Her background is product: PM at Asana (2021-2023) shipping revenue-generating features, Product Lead at Apella (healthcare AI, 2023-2024), and she's explicitly still 'looking to build my founding team' as of LinkedIn. No co-founder listed with engineering or ML credentials, which is a red flag for a company claiming to do post-training research.

Low Signal

Competition

Direct competitors include Scale AI, Surge AI, Appen, and Labelbox — all well-funded and established in the training data space. Scale AI in particular has deep relationships with OpenAI, Google, and Meta. Rubric's differentiation angle (credentialed expert judgment, RL environments for non-verifiable domains) is intellectually interesting but unproven against incumbents with massive head starts and distribution.

Medium Signal

Product

Website describes a full-stack post-training platform including expert network tooling, RL environments, and multi-modal data infrastructure. Claims to be 'trusted by' frontier AI labs, enterprise healthcare, legal tech, and financial services — but no named customer logos, no pricing, no metrics, no case studies. 'Talk to Us' CTA with no live demo or API docs visible.

OverallC Tier

Rubric AI is attacking a real and large market with a compelling framing around 'non-verifiable knowledge' and expert-grounded RL environments — but the execution risks are severe. The solo non-technical founder is a critical red flag for a company positioning itself as a post-training research lab; Pragya is still actively recruiting her founding team, meaning the core technical product may not yet exist in meaningful form. No named customers, no metrics, no co-founder with ML/engineering credentials, and direct competition from Scale AI with its billions in revenue and entrenched lab relationships. The idea is differentiated on paper but needs a strong technical co-founder and evidence of actual contracted revenue to be credible.

Active Founders

Pragya Saboo

Founder

cofounder at Rubric AI; previously product at Asana, Oscar Health, and Apella

Rubric AI

TierC Tier

BatchWinter 2026

Team Size2

StatusActive

LocationSan Francisco, CA, USA