Three companies. Same move.
In the last six months, three of the most-watched SaaS companies answered “bundle AI or charge separately?” the same way: kill the standalone AI tier, bundle it into a higher plan, raise the higher plan’s price.
Notion
+33% on Business. AI bundled in. Standalone AI tier retired.
Slack
+25% on Business+. AI moved into the tier. Add-on closed.
Loom
AI rolled into Business tier. Standalone retired.
Prices went up, not down — even though the AI is now “included.”
Why bundling beats unbundling for AI
The standalone AI tier was the safe answer. It let you charge the customers who valued AI, while leaving the existing base untouched. The problem is what it implicitly tells the market: AI is a separate product, valued separately, priced separately.
That framing undersells what AI does to the whole product. When AI lands inside the core experience — smarter search, faster drafting, fewer manual handoffs — it raises the perceived value of everything, not just the AI feature. Charging separately leaves that lift on the table.
The ARPU math agrees. A 25–35% list-price bump on the core plan, with AI bundled in, beats a standalone $10–20 AI add-on at typical attach rates by a comfortable margin. And it does it without forcing the customer to negotiate “is the AI tier worth it?” every cycle.
When not to bundle
Bundling breaks down in two cases.
Usage-based AI products. If the AI feature scales linearly with compute cost — image generation, long-context inference, agentic workflows — bundling it at a flat price exposes you to unbounded margin compression. The right answer is a usage-metered AI line item, not a separate tier.
Buyer-misaligned AI. If the people who value AI inside your product are systematically different from the people who buy your product (often: practitioners vs procurement), bundling forces buyers to pay for something their stakeholders won’t credit them for. A premium add-on the practitioner can pull in works better.
Everywhere else, bundle.
