The signal this week is not a funding round. It is a price tag. When the largest software incumbents on earth start charging per result instead of per user, they are telling you something about the unit of value that the rest of the market has not fully priced in yet.

The Signal

The moves have stacked up fast. Intercom’s Fin agent charges $0.99 per resolved conversation, with no platform fee — you pay only when the agent actually closes the ticket. Salesforce, after two years of pricing whiplash, now runs three Agentforce models simultaneously: $2 per customer-facing conversation, $0.10 per action via flex credits, and $125 per user per month for internal agents. And in April 2026, HubSpot switched its AI pricing from per-use to per-resolution — landing near $0.50 a resolution, undercutting Intercom and turning outcome pricing into a competitive weapon rather than an experiment.

This is not three vendors testing a clever billing scheme. It is a regime change. By April 2026 the pattern spanned the category: Sierra pricing purely on outcomes, Intercom scaling its per-resolution model, Salesforce hedging across three pricing structures at once, and Zendesk already converted. The demand side confirms it: Futurum’s first-half 2026 buyer survey found 43% of buyers now prefer consumption-based pricing and 27% favour outcome-based structures, with hybrid models — a fixed base plus variable usage — emerging as the dominant transition state.

The underlying cause is mechanical, not fashionable. AI agents do not occupy seats. They do not log in, accumulate a user profile, or appear on a license dashboard; they execute work — sometimes thousands of tasks — without consuming a single credential. A pricing model built to count humans cannot meter a workforce that has no headcount. As one analysis put it, per-seat pricing is being punished by markets precisely because the seat has stopped corresponding to value delivered.

Why It Matters

For founders building agentic products, this is permission and pressure at once. Permission, because the biggest incumbents have now validated outcome pricing — you no longer have to educate the market that “charge per resolution” is legitimate. Pressure, because a competitor pricing on outcomes can credibly say “you only pay when it works,” and a per-seat incumbent cannot match that without cannibalising its own revenue model. The companies most exposed are the mid-tier SaaS vendors whose entire business is seat counts that AI is about to compress.

For operators buying these tools, the renewal math is changing under you. The right question at your next renewal is no longer “how many seats do we need?” but “what does a unit of outcome cost, and can we measure it?” Outcome pricing only protects the buyer if the outcome is defined tightly enough that the vendor cannot inflate it.

The Charaka View

We have argued from the start that the seat is the wrong unit for an agentic world — our own architecture is priced and reasoned in agents and tasks, not headcount, because that is what AI-native operations actually consume. The signal worth detecting here is second-order: when pricing moves to outcomes, it forces vendors to expose whether their agents work, because you can no longer hide an unreliable product behind an annual per-seat contract. Outcome pricing is, quietly, an accountability mechanism — and that is exactly why the incumbents resisted it for two years and why the ones with confident, measurable agents are now leading with it.

This analysis draws on SaaStr’s reporting on Salesforce’s three Agentforce pricing models, SaaStr’s coverage of HubSpot’s switch to per-resolution pricing (Apr 2026), Fin.ai’s per-resolution vs per-conversation analysis, Futurum Group’s 1H 2026 pricing survey, and Monetizely’s analysis of Agentforce pricing. Human editorial oversight applied.

This analysis is informational and does not constitute investment advice, a research report, or a recommendation to buy, sell, or hold any security.

Charaka Notes by Manthan Intelligence. Subscribe