a16z Backs Vals: Bringing Real-World Workflow Benchmarks to AI Evaluation
a16z Newsletter · rss · 2026-08-13
a16z has announced its investment in Vals, an AI model evaluation startup. The article notes that as frontier models saturate traditional academic benchmarks, model capabilities are shifting from "answering questions" to "doing the work" (e.g., handling complex enterprise workflows, running agents autonomously for hours).
Vals aims to be the "independent scorekeeper" for the AI market. Its core approach transforms real-world workflows from domain experts into rigorous benchmarks, building automated grading systems. For instance, instead of testing if a legal model can pass the bar exam, Vals tests if it can perform actual legal research. Additionally, Vals has built scalable, contamination-resistant infrastructure that delivers results within hours and continuously retires saturated tests to introduce new tasks.
More from Venture
- Dev builds MCP server that lets AI agents outsource phone calls to real humans — Many_Benefit_3393 · 2026-10-16
- Jensen Huang: Nvidia charges ~$thousands per GPU yearly for software, a major revenue line — rohanpaul_ai · 2026-10-16
- Creator shares the agent-powered system behind millions of monthly impressions — VibeMarketer_ · 2026-10-16
- Skip the $49 SaaS: Mid-size businesses will pay $5K-10K for someone who just implements AI — aryanXmahajan · 2026-10-16
- Indie site built over a holiday earns a Yahoo writeup and a Wikipedia backlink — gefei55 · 2026-10-16
- New startup path: VC-funded founders become creators, then incubate products solo — FinanceYF5 · 2026-10-16