Traditional SaaS switching costs were based on painful data migrations, which LLMs may now automate. The new moat for AI companies is creating deep, customized integrations into a customer's unique operational workflows. This is achieved through long, hands-on pilot periods that make the AI solution indispensable and hard to replace.

As AI commoditizes user interfaces, enduring value will reside in the backend systems that are the authoritative source of data (e.g., payroll, financial records). These 'systems of record' are sticky due to regulation, business process integration, and high switching costs.

As AI and better tools commoditize software creation, traditional technology moats are shrinking. The new defensible advantages are forms of liquidity: aggregated data, marketplace activity, or social interactions. These network effects are harder for competitors to replicate than code or features.

AI capabilities offer strong differentiation against human alternatives. However, this is not a sustainable moat against competitors who can use the same AI models. Lasting defensibility still comes from traditional moats like workflow integration and network effects.

Satya Nadella predicts that SaaS disruption from AI will hit "high ARPU, low usage" companies hardest. He argues that products like Microsoft 365, with their high usage and low average revenue per user (ARPU), create a constant stream of data. This data graph is crucial for grounding AI agents, creating a defensive moat.

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.

Point-solution SaaS products are at a massive disadvantage in the age of AI because they lack the broad, integrated dataset needed to power effective features. Bundled platforms that 'own the mine' of data are best positioned to win, as AI can perform magic when it has access to a rich, semantic data layer.

In enterprise AI, competitive advantage comes less from the underlying model and more from the surrounding software. Features like versioning, analytics, integrations, and orchestration systems are critical for enterprise adoption and create stickiness that models alone cannot.

The existential threat from large language models is greatest for apps that are essentially single-feature utilities (e.g., a keyword recommender). Complex SaaS products that solve a multifaceted "job to be done," like a CRM or error monitoring tool, are far less likely to be fully replaced.