Sierra hit $150 million in annualized revenue eight quarters after launch by charging for outcomes instead of seats. The per-seat pricing model fell from 21 percent to 15 percent of SaaS vendors in twelve months. Companies still pricing AI agents per seat are subsidizing their own disruption.
In March, this journal predicted that outcome-based pricing would account for more than 25 percent of new enterprise AI agent contracts by the end of 2026. That prediction is tracking ahead of schedule. The evidence is no longer directional — it is structural.
Sierra hit $150 million in annualized revenue in February 2026, eight quarters after Bret Taylor and Clay Bavor launched it. The company raised $950 million at a $15.8 billion valuation in May. It serves 40 percent of the Fortune 50. None of this is remarkable for a well-funded AI startup. What is remarkable is the pricing model that produced these numbers: Sierra charges its customers for outcomes — problems resolved, tasks completed, questions answered — not for seats occupied.
The Price List
The competitive landscape has converged on a unit of measurement that did not exist eighteen months ago: the automated resolution. Intercom charges $0.99 per resolution through its Fin agent. HubSpot cut its Customer Agent price from $1.00 to $0.50 per resolved conversation in April. Zendesk charges $1.50 on committed volume and $2.00 for pay-as-you-go overages. Each vendor has independently arrived at the same conclusion — the customer should pay for what the agent accomplishes, not for the right to access it.
The per-seat model that built enterprise software into a $2 trillion industry is contracting. Growth Unhinged's 2025 State of B2B Monetization report shows seat-based pricing fell from 21 percent to 15 percent of SaaS vendors in twelve months. Hybrid models — combining a platform fee with usage or outcome charges — surged from 27 percent to 41 percent over the same period. The remaining vendors split between pure usage-based and outcome-based approaches. The direction is unambiguous.
This is not a philosophical preference. It is an arithmetic inevitability. When an AI agent handles a customer support inquiry, the vendor incurs compute cost proportional to the work performed. A per-seat model charges a fixed fee regardless of how many inquiries the agent handles. As agents improve and handle more, the vendor's compute costs rise while revenue stays flat. The margin compresses with every capability gain. Outcome-based pricing reverses this: revenue scales with the work the agent performs, aligning cost and income on the same axis.
The Tax
The outcome tax is what per-seat companies pay for having the wrong pricing model during an AI transition.
Consider two identical customer support platforms. Platform A charges $50 per seat per month. Platform B charges $0.99 per resolution. Both deploy AI agents that can handle routine inquiries. In a traditional environment with human agents handling 200 tickets per month per seat, Platform A earns $50 and Platform B earns $198. Platform B earns more, but the difference is manageable.
Now the AI agent improves. It handles 500 tickets per month. Platform A still earns $50 — the customer is paying for the seat, not the work. Platform B earns $495. The gap has widened from $148 to $445. The customer on Platform A is getting more value without paying more. The customer on Platform B is paying more because they are receiving more. Platform A's margin compresses. Platform B's expands.
This is the structural mechanism beneath the pricing data. As AI agents improve — and they are improving on a quarterly cadence — every per-seat contract becomes a subsidy. The vendor absorbs rising compute costs while the customer captures rising capability at a fixed price. Outcome-based vendors capture value proportional to the capability they deliver. The gap compounds with every model upgrade.
Sierra's trajectory illustrates the compounding. The company crossed $100 million in ARR in November 2025 and $150 million three months later — a 50 percent increase in a single quarter. Outcome-based revenue accelerates as the underlying agents handle more, because each additional resolution generates incremental revenue. Per-seat revenue does not accelerate. It is structurally flat.
Winners and Losers
The winners are companies that built outcome-based pricing before the transition forced their hand. Sierra never sold seats. Intercom rebuilt its entire pricing model around Fin's per-resolution economics. HubSpot's April price cut from $1.00 to $0.50 per resolution was a competitive move — aggressive enough to undercut Intercom and Zendesk, cheap enough to drive adoption volumes that make up the difference at scale.
The losers are companies whose revenue depends on seat count in categories where AI agents are replacing seated humans. Freshworks generates the majority of its customer support revenue from per-seat licenses. Zendesk has introduced outcome-based pricing for its AI agent tier but still relies heavily on per-seat revenue for its core platform. Both face the same structural pressure: their best customers — the ones whose agents handle the most inquiries — are the ones generating the least incremental revenue.
Salesforce sits in the middle. Agentforce crossed $1.2 billion in ARR in May, up 205 percent year over year. The $2 per conversation charge is outcome-adjacent, but the core platform remains seat-based. Salesforce is running two pricing models simultaneously — a per-seat legacy that built a $35 billion business and a per-outcome experiment that is growing faster than anything else in the portfolio. The transition is half-built.
The Falsifiable Claim
If more than 50 percent of new enterprise AI agent contracts are still priced on a pure per-seat basis by Q4 2026, this thesis is wrong. The per-seat model would have demonstrated resilience that the current trajectory does not suggest. The Growth Unhinged data shows it at 15 percent and falling. But enterprise procurement moves slowly, and incumbents have strong incentives to preserve the pricing model that built their installed base. The question is whether the structural advantage of outcome-based pricing is large enough to overcome institutional inertia within two quarters.
The evidence from Sierra, Intercom, and HubSpot suggests it is. The evidence from Freshworks and legacy Zendesk suggests the inertia is real. The outcome is determined by which force compounds faster — the economic logic of paying for outcomes, or the procurement friction of changing how enterprise budgets are structured.
The per-seat model built enterprise software. It will not build the agent economy. The companies that recognize this early are compounding. The companies that do not are paying the outcome tax — subsidizing AI capabilities they cannot charge for, one improved resolution at a time.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)