a16z: Vertical AI startups still have whitespace — four traits of a winnable market
iamrobotbear · x · 2026-09-04
a16z partner Seema Amble lays out four criteria for picking a market where vertical AI startups can win: work repeats often enough to learn from; judgment matters so there's nuance to learn; an expert can quickly evaluate output and explain improvements; and you can start with one task and expand to owning more of the job. Together these form a strong learning loop — the core vertical AI advantage.
Framed against the Claudeforce debate (Salesforce + Anthropic letting Claude operate Salesforce CRM), she concedes incumbents are moving beyond chatbots — Docusign Iris reviews contracts, Atlassian Rovo routes requests, Klaviyo Composer builds campaigns. But incumbents own the record, labs own the front door, and vertical AI can still own the job via data assets, depth, learning loops and cross-system coordination.
Related event: a16z: Incumbents Go Agentic, but Vertical AI Startups Still Have Openings(3 posts)→
More from Venture
- Daytona raises $60M Series B at $1B valuation to build the agent cloud — tsi_org · 2026-10-21
- Satirical Post: $1B 'AI Trading Lab' Allegedly Trades on Model-Release Insider Info — StewartalsopIII · 2026-10-20
- a16z's Ben Horowitz says AI buildout isn't the 90s bubble — an investor pushes back — henloitsjoyce · 2026-10-20
- DeepSeek hands inference providers 90%+ margins, pressuring Moonshot and Zhipu hosting deals — teortaxesTex · 2026-10-20
- a16z leads Eka Robotics Series A, betting on Vision-Force-Action models for dexterous manipulation — a16z · 2026-10-20
- Salesforce signs definitive agreement to acquire customer engagement platform Braze — thedealdirector · 2026-10-20