Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
.png)
Placing our five companies on those three dimensions looks like this:
Company Zero Human Domain Output/Cost Cursor M M Inflecting Devin L M Inflecting to Exponential Harvey AI M L Exponential 11x (Alice) L M Inflecting Sierra AI L L Exponential
Figure 6. The Agentic Monetization Spectrum - as autonomy, domain breadth, and the output-to-cost curve increase, the right pricing metric moves from per-seat toward outcome.
Each company’s position points to the pricing metric it should use. Here is that read, company by company, with our take on whether the current metric fits. (see Figure 6)
Cursor (M / M / Inflecting): The human is still the quality gate, and the domain is a single function. Per-seat fits naturally. The buyer is paying for productivity gains per developer.
Current metric: Per seat. Our assessment: Correct for its AMS position. The risk is that Cursor’s agent grows beyond per-seat before the market is ready for something more complex. But getting the structure right matters more than the exact price, and per-seat gives Cursor a clean, predictable base to build on.
Devin (L / M / Inflecting to Exponential): The agent does the work, but still inside one function. This is where per-seat starts to feel like a ceiling and per-output starts to look possible. Cognition chose usage. The ACU model is a hybrid: part platform fee, part usage.
Anchoring ACUs to compute cost instead of output value is deliberate. Devin succeeds on only 15 to 30% of complex tasks, so value-based pricing would invite ROI scrutiny its reliability cannot yet survive. As the success rate climbs, Cognition has a clear path to add output-quality tiers (a "best effort" ACU versus a "guaranteed merge-ready" ACU at a premium) without changing the metric.
Current metric: Hybrid, platform fee plus usage (ACUs). Our assessment: One of the best-designed pricing metrics in agentic AI. The base captures the value of "access to an autonomous worker," while ACU usage scales with the actual work done.
Harvey AI (M / L / Exponential): The output dwarfs the cost, so the Output/Cost Curve points clearly to value-based pricing. Yet Harvey uses per-seat. This looks like a mismatch: the seat model earns the same whether a lawyer uses Harvey once a week or twenty times a day. But Harvey’s per-seat metric is more defensible than it looks. Harvey’s buyers are law firms whose whole business runs on billable hours and seat-based software. They budget by headcount. Outcome pricing might be better economically, but it creates procurement friction in an industry that does not buy that way.
Current metric: Per seat. Our assessment: Not ideal for the AMS position, but defensible given how law firms buy. Harvey should evolve by adding a usage layer on top: tiered pricing where heavy-usage firms pay a premium, or a "matter-based" add-on for high-volume due diligence.
11x / Alice (L / M / Inflecting): The agent runs on its own, but output quality is uneven enough that the real value ratio is not reliably high. Reports suggest Alice needs 10,000 or more emails to produce meaningful results.
The flat monthly fee creates misaligned incentives. If Alice underperforms, the buyer feels overcharged. If Alice overperforms, 11x earns the same either way: the gym-membership model. It only holds up as long as the buyer is not watching ROI closely.
Current metric: Flat monthly fee. Our assessment: A hybrid model (a lower base plus a per-qualified-meeting kicker) would align incentives and let 11x earn more from high-performing deployments. Competitive pressure makes pure value-based pricing risky, but a modest outcome component would set 11x apart from the commodity players racing to zero.
Sierra AI (L / L / Exponential): Fully autonomous, cross-functional, with an output-to-cost ratio that keeps widening. A resolved customer interaction costs pennies in compute. The value of that resolution (avoiding a $5 to $15 human ticket, keeping a $500-a-year subscriber, upselling a product) far outpaces the cost. Sierra charges per successful resolution. This is the best-aligned pricing metric in the whole set. It ties directly to client value, it is measurable without dispute, and it creates real incentive alignment.
Current metric: Per outcome. Our assessment: Correct for its AMS position. Outcome-based pricing is not a goal every agent should chase. It works only when specific conditions are met: clear attribution, immediate measurement, and high value per outcome. Sierra meets all three.
Join companies like Zoom, DocuSign, and Twilio using our systematic pricing approach to increase revenue by 12-40% year-over-year.