AfterQuery, a San Francisco-based provider of expert-generated training data for AI models, has raised $30M in Series A funding led by Altos Ventures. The company curates datasets, benchmarks, and RL environments from 100,000+ verified professionals in domains like finance, medicine, and law to enable LLMs and agents to handle complex reasoning tasks. The capital will expand its expert network, deepen domain coverage, and support hiring.
Data Scarcity Drives Expert Curation Surge
The raise arrives amid acute data bottlenecks for frontier AI labs, with Protege securing a $30M Series A extension in January 2026 for real-world multimodal data. Scale AI holds a $13.8B valuation after a $1B Series F in 2024, while Surge AI bootstrapped to over $1B ARR. AfterQuery differentiates by encoding tacit expert knowledge into reasoning traces and agent environments, powering every major US frontier AI lab.
Frontier Models Exhaust Public Data
AI training dataset market stands at $2.82B in 2024, projected to reach $9B by 2029 per MarketsandMarkets. Public web data fails to capture nuanced professional decisions, tradeoffs, and context, leading to hallucinations in agentic tasks like quant trading or IDE navigation. Labs like Anthropic and OpenAI now rely on specialized providers to scale next-gen models.
Encoding Expertise into RL Environments
AfterQuery draws from its network of nearly 100k verified practitioners to produce custom datasets for SFT, RLHF, and multimodal tasks, alongside benchmarks like IDE-Bench and Terminal-Bench. Unlike Scale AI's broad annotation platform or Snorkel AI's programmatic labeling, AfterQuery focuses on human-demonstrated chain-of-thought for real workflows. Its internal post-training pipelines validate data quality by boosting model scores over 5x on Terminal-Bench 2.0.
As Spencer Mateega, Co-Founder & CEO, noted:
"We have researchers internally create a post-training pipeline… objectively show… data’s high quality."
Altos Backs Rapid AI Data Scale-Up
Altos Ventures led the round, joined by The Raine Group, Y Combinator, BoxGroup, and angels from Anthropic, OpenAI, DeepMind, Meta, and Microsoft. This investor mix signals conviction in AfterQuery's researcher-partner model over generic labeling. The $300M valuation reflects $100M ARR achieved in 14 months post-YC W25.
Benchmarks Expose Agent Gaps
AfterQuery publishes open evals like FinanceQA, Market-Bench, and VADER to highlight LLM shortcomings in professional domains. Competitors like Labelbox ($250M+ raised) emphasize annotation tools, while AfterQuery delivers ready-to-use RL setups from real codebases. Rising agentic AI demand shifts capital toward expert signals amid 27.7% CAGR market growth.
YC Alumni Founders Hit $100M ARR
Co-founders Spencer Mateega (ex-Google, Meta; Wharton), Carlos Georgescu (ex-Citadel, Meta; prior ed-tech exit), and Danny Tang (Wharton; prior startup sale) launched via YC Winter 2025. Their FAANG and finance pedigrees enable rapid execution, growing from three founders to 51-200 employees serving Fortune-scale AI labs. Previous high school startup exit underscores serial entrepreneurship.
Expert Network Expansion Ahead
Funds target hiring across engineering, research, and operations, plus scaling domain experts for new verticals. Recent calls seek AI/ML interns and finance professionals, building on partnerships like The Raine Group. With deployments proving 5x benchmark gains, AfterQuery eyes deeper integration into global AI workflows.