a16z: three rules for how AI applications earn their premium
a16z Newsletter · rss · 2026-11-02
a16z Newsletter analyzes pricing and moats for AI application companies: falling inference costs help margins, but stronger models and easier dev tools make core features easier to replicate. The key question is what customers will keep paying for.
Cases:
- Repsol legal: blind-tested Harvey vs Microsoft Copilot vs its internal prototype Lexia; Harvey won on quality, then beat alternatives in an 8-week pilot across 12 jurisdictions with 50 lawyers. Full rollout followed — lawyers save 4-6 hours weekly, 96% adoption, and Harvey-derived insights improved contract templates.
- Notion × Decagon: after weighing a build, Notion chose Decagon; ticket resolution time improved up to 34%.
- A vertical software company with proprietary pricing benchmarks across 3,000 companies is exploring renting that context via an MCP server.
Three rules:
- Operate the workflow better than the customer can — one in-housed high-volume task is enough to erode your premium; win across the full workflow (standards application, edit consistency, context preservation).
- Make context cheaper to rent than to rebuild — collect, validate, maintain once; spread across customers.
- Prove value, then price to capture it — in a16z's Aug 2026 survey of 100 enterprise LLM decision-makers, 94% expected LLM spending to rise and 37% expected it to at least double within a year. Price per recognizable unit of completed work, including retries and human review.
More from Venture
- Meta's Muse for SMBs: 500M businesses, and 25% of users already shop via the agent — omooretweets · 2026-11-03
- Northwood raises $375M Series C to scale Earth-space data connectivity — oyhsu · 2026-11-03
- Anthropic Grows Revenue Twelvefold but Posts $23 Billion Operating Loss — The Decoder · 2026-11-03
- Pathfinder raises $16M Series A for ESA school-choice fintech — darian314 · 2026-11-03
- Indie Dev's Xiaohongshu Followers Jump to 7,150 in About a Week — ezshine · 2026-11-03
- Tutor Intelligence raises $75M Series B led by a16z to deploy teleop-trained robots — a16z · 2026-11-03