The Per-Seat Pricing Model Is Dying in Real Time Across Three Unrelated Categories

Three companies in three completely different SaaS categories arrived at the same pricing architecture in Q1 2026, and I don't think it's a coincidence.

ServiceNow introduced Pro Plus premium tiers at 25-45% above standard pricing for autonomous AI capabilities. Their Now Assist product hit $1B ACV run rate, the fastest product launch in company history. Salesforce shifted Agentforce to Flex Credits at $0.10 per action or $2 per conversation, explicitly decoupling revenue from human headcount. Adobe implemented Generative Credits for AI features.

All three now bill for AI agent work output rather than human seats. Hybrid pricing (base fee plus consumption) now covers 43% of SaaS companies, projected to hit 61% by end of 2026.

This connects to a deeper structural problem: seat compression. Atlassian reported its first-ever enterprise seat count decline and cut 1,600 jobs. MongoDB is down 42% as vibe coding reduces the number of developers per project. ServiceNow dropped nearly 50% from its 2025 peak on fears that AI agents eliminate the support tickets driving seat purchases. If one AI agent replaces 10 knowledge workers, that's 10 fewer seats across every SaaS tool those workers used.

For builders pricing their own products: watch this carefully. SaaStr's Jason Lemkin argues that AI agent configurations are fundamentally portable (prompts, tools, instructions), meaning the switching costs that protected SaaS companies don't protect AI-native platforms. The moat isn't the model or the workflow. It's the data underneath. ServiceNow's strongest asset isn't ticket processing, it's the operational record. Canva's stealth churn problem is power users replacing individual workflows with AI specialists while still appearing as healthy customers in NPS.

If you're selling per-seat in 2026, you're selling a melting asset. The market has decided.