How should you price an AI SaaS product?

Founders and investors agree that AI has shifted the software industry from the Cloud Era, defined by seat-based subscriptions for data access, to the AI/Value Era, where software is priced based on the work it performs a16z — Tech Executives: AI Has Changed SaaS Forever @ 06:06.

The "Golden Quadrant": Outcome-Based Pricing

For AI products that are both highly autonomous (requiring little human intervention) and have high "attribution" (clear, measurable results), experts recommend outcome-based pricing Lenny's Podcast — Pricing your AI product @ 39:41.
* Capturing Value: While traditional SaaS typically captures 10–20% of customer value, autonomous AI can capture 25–50% because it competes with labor budgets rather than software budgets Lenny's Podcast — Pricing your AI product @ 42:45.
* Examples: Intercom’s Finn charges $0.99 per successful automated resolution, while Sierra prices based on "call deflections" in customer service Lenny's Podcast — Pricing your AI product @ 48:52, Lenny's Podcast — He saved OpenAI: Bret Taylor (Sierra) @ 1:02:55.
* The Labor Budget Trap: If you use old playbooks to price "agentic" products, you risk "under-monetizing from day one" by charging software rates for tools that replace expensive human labor Lenny's Podcast — Pricing your AI product @ 27:29.

The Hybrid Model: Bridging Seats and Usage

For many startups, a hybrid model is currently the most popular archetype. This retains a base seat-based fee for human "co-pilots" while layering in usage-based credits or tokens for AI-driven tasks Lenny's Podcast — Pricing your AI product @ 39:41.
* Predictability: Large enterprise buyers often prefer seat-based pricing because it provides budget predictability; OpenAI’s ChatGPT Enterprise still uses seats for this reason 20VC with Harry Stebbings — Gokul Rajaram on the 8 Moats @ 30:30.
* Expansion: While seats provide a floor, usage-based components drive expansion revenue as customers scale their AI adoption 20VC with Harry Stebbings — Gokul Rajaram on the 8 Moats @ 30:30.

Strategic Considerations for Founders

• Don't Price on Margin: a16z advises against pricing based on your GPU costs ("cost-plus"). Instead, price against the status quo or the cost of the human work you are augmenting or replacing a16z — What Is an AI Agent? @ 18:16.
• Usage vs. "Casino Tokens": Some warn that customers may "hate" consumption-based models if they feel out of control or confused by "casino chips" (arbitrary credit systems) a16z — Atlassian CEO on the SaaS Apocalypse @ 30:32.
• ROI-Led Sales: In the AI era, the sales cycle can be shortened if you drop in a tool that "pays for itself in the same month," making it look more like a service than a traditional software purchase Y Combinator — AI Revolution: What Nobody Else Is Seeing @ 30:31.

Where they disagree:
While many push for outcome-based models, HubSpot CEO Yamini Rangan argues that for platforms, AI should be "inbuilt" into the current product rather than a separate premium "AI SKU," focusing on overall platform value rather than isolated AI outcomes 20VC with Harry Stebbings — Yamini Rangan: How HubSpot Competes @ 03:03. Conversely, Notion is cited as a rare success that successfully doubled its seat price by adding an AI layer 20VC with Harry Stebbings — Predictions for 2026 @ 48:52.

— Sources: 11 videos across 4 creators