Notion's Third-Party AI Dependency

This turns Notion’s AI upsell into a margin and control problem, not just a product feature story. Since May 2025, unlimited AI has been bundled into Business and Enterprise, so the clearest path to larger contracts now runs through features built on outside models. That helps Notion sell bigger seats and richer workflows like search, meeting notes, and drafting, but it also means a meaningful piece of enterprise value depends on model vendors that set the underlying cost, quality, and roadmap.

• The packaging makes the dependency more important than before. Notion no longer sells AI as a simple add on for most customers. It uses AI to pull teams up into higher tiers, which means vendor powered features are tied directly to contract expansion and ARPU growth.
  1 notion 2 sacra
• Notion’s job is orchestration, not model building. It stitches AI into page editing, search across connected apps, meeting notes, and database workflows, while relying on providers like Anthropic and OpenAI for the underlying reasoning and generation. That keeps R&D lighter, but leaves core capability outside Notion’s control.
  2 sacra 5 sacra 6 anthropic
• This is especially sensitive in enterprise because incumbents bundle their own AI into broader suites. Microsoft can include Loop access inside existing E3 and E5 estates, while Notion has to buy intelligence upstream and resell it downstream. That creates more pressure on gross margin and pricing discipline as AI usage grows.
  4 microsoft 7 sacra

The next step is deeper vertical integration around workflows, connectors, and agents, so customers buy Notion for getting work done inside their stack, not just for raw model output. The more Notion can wrap third party models in proprietary context, approvals, and multi app automation, the more durable its enterprise pricing becomes.