Software that is priced per seat and easy to replace, like Zendesk for customer support, is under existential threat from AI. Customers can run AI agents in parallel to perform the same tasks, directly comparing performance and cost, making it easy to reduce seats and switch providers.

Traditional SaaS companies are trapped by their per-seat pricing model. Their own AI agents, if successful, would reduce the number of human seats needed, cannibalizing their core revenue. AI-native startups exploit this by using value-based pricing (e.g., tasks completed), aligning their success with customer automation goals.

Investor Joe Lonsdale offers a heuristic for the 'SaaSpocalypse': low-end SaaS, particularly PE-backed companies that prioritized sales over deep tech, is in trouble. However, complex software that required over $100 million in engineering to build has a significant moat and is defensible against AI-driven disruption for the foreseeable future.

The fundamental business model of many SaaS companies is based on per-user pricing. AI agents pose an existential threat to this model by enabling smaller teams to achieve the same output as larger ones. As companies wonder why they should pay for 100 seats when 10 people can do the work, the entire economic foundation of the SaaS industry faces a crisis.

AI agents can easily siphon off value from SaaS products priced on per-seat utility by automating tasks previously done by humans (e.g., support tickets). In contrast, deeply embedded systems of record (ERP, CRM) are insulated by career-limiting switching costs and the immense challenge of migrating timeless, critical data.

The market has overreacted to AI's threat to SaaS giants like Salesforce and Adobe. While AI can replicate code, it cannot easily replace the years of deep integration into client billing, customer service, and employee training. These high switching costs are being ignored, making their stocks undervalued.

SaaS business models derive value from long-term customer relationships. AI's disruptive potential makes the 10-year outlook for any software company extremely uncertain. This means the entire SaaS category is currently mispriced, though it's unclear if companies are over or undervalued.

Sierra CEO Bret Taylor argues that transitioning from per-seat software licensing to value-based AI agents is a business model disruption, not just a technological one. Public companies struggle to navigate this shift as it creates a 'trough of despair' in quarterly earnings, threatening their core revenue before the new model matures.

As AI agents perform more work and human headcount decreases, the traditional seat-based pricing model becomes obsolete. The value is no longer tied to human users. SaaS companies must transition to consumption-based models that charge for the automated work performed and value generated by AI.