Sierra, the AI customer service company built to replace live agents with a system that only gets paid when it resolves a ticket, closed a $950 million round led by Tiger Global and GV at a $15.8 billion valuation on May 4, 2026. The company does not sell seats. It sells resolved outcomes, and investors just priced that model at nearly $16 billion.
Sierra is not an outlier. On January 28, 2026, Decagon raised a $250 million Series D led by Coatue Management and Index Ventures that tripled its valuation to $4.5 billion, according to Bloomberg. Decagon prices entirely on usage, per conversation or per resolution, with no seat count anywhere in the contract.
The growth gap behind the money
The capital is following a measurable growth difference, not a hunch. Bessemer Venture Partners' AI pricing and monetization playbook, published February 9, 2026, found that AI companies tying pricing to customer outcomes grow 1.5x to 3x faster than peers still charging per seat.
The broader market is already moving to match that data. A survey of 240 software and AI companies, run April through May 2025 by Kyle Poyar's Growth Unhinged, found seat based pricing fell from 21 percent to 15 percent of companies in twelve months, while hybrid pricing combining a subscription with usage climbed from 27 percent to 41 percent, published June 4, 2025. Hybrid, not pure usage, is the majority path. Most vendors are not ripping out subscriptions. They are bolting a usage meter onto them.
Incumbents are hedging, not holding the line
Even vendors built on seats are building an exit ramp. Salesforce now sells its Agentforce platform on a consumption model called Flex Credits, priced at $500 per 100,000 credits, about $0.10 per standard agent action, alongside its older $2 per conversation and $5 per user per month options. Intercom's Fin AI Agent skips seats entirely. It charges $0.99 for each billable outcome, a resolution, a procedure handoff, or a disqualification, and $9.99 per qualification, with no per seat licensing required to deploy it.
Vertical software is following the same path with a different mechanic: bundled, case based pricing instead of metered credits. On May 14, 2025, EvenUp, which sells AI tools to personal injury law firms, replaced its per document charges with all in one, case based pricing as part of its AI Drafts Suite and Smart Workflows launch. A law firm does not buy a seat or a document count anymore. It buys a case handled.
The accounting is catching up, which means the model is real
The clearest sign this is structural and not a pricing fad: the accountants showed up. On June 4, 2026, Deloitte published formal guidance for agentic AI vendors on how to recognize revenue when a company charges only for successful outcomes, like a resolved support ticket, rather than for seats, distinguishing a stand ready access obligation from an output based delivery obligation. Standards bodies do not write guidance for pricing models that are still a niche experiment. They write it because CFOs are already asking how to book the revenue.
Sierra, Decagon, Salesforce, and Intercom did not abandon pricing certainty for growth. They found customers will pay more, faster, for a price tied to the result they actually wanted.
The operator takeaway is specific, not directional. If your vertical AI product still bills per seat, price a second column this week: a per outcome or per resolution rate for the same workflow, and offer it to three existing customers as an option, not a replacement. Watch which ones self select into it. That is your signal for whether the whole book should move, and the accounting guidance now exists to help you book it correctly once it does.