In May 2025, Notion stopped selling its $8/seat AI add-on to new customers. AI moved into the Business tier ($18/seat) only. Plus stayed at $10. New teams who wanted AI had to go from $10 to $18, a roughly 80% lift. The line on the bill called it "bundling." The line on the internal Notion strategy doc almost certainly called it something else.
Slack followed in 2026. Loom is on the same path. The pattern matters because it's now the default playbook for monetising AI features without printing a price increase.
The mechanic
The standalone AI add-on, at $8 to $10 per seat per month across SaaS, was always a transitional product. Vendors needed a way to charge for AI without making it look like a price hike. The add-on was the polite version: "we built something new, here's the optional surcharge."
The polite version had a problem: most users bought into the floor tier and skipped the add-on. AI revenue per seat was lower than the model providers charged for the underlying tokens. The economics didn't work.
So vendors switched to bundling. Move AI into a higher base tier, force users who want AI to migrate up, and pocket the spread. The headline price didn't change for users who stayed where they were. But the path to use the new feature now costs $8 to $10 more per seat than it did six months ago.
Bundling is a price increase that doesn't show up in the headline number. If you're an existing customer, you don't notice. If you're a new customer, your floor for the same effective product moved up.
Three tells that bundling is happening
Watch for these three signs in your tools. They're how the bundling move shows up in the wild.
1. The add-on becomes "unavailable for new customers"
This is the tell. Notion's AI add-on still exists, but only for accounts that subscribed to it before May 13, 2025. Anyone signing up today on Plus literally cannot buy AI. The fine print uses words like "this feature is now included in Business" but the practical effect is that the cheap path to AI is closed.
If you're grandfathered, you keep the deal. The catch is that any cancellation, even for a single billing cycle, ends the grandfather clause. You don't get to come back at the old rate.
2. The "no price change" announcement
The vendor blog post will say "no change to existing prices." Read closely. What didn't change is the headline price of each tier. What did change is which features sit in which tier.
The most aggressive version of this, in 2026, has been Jira Cloud's October 2025 hike (5 to 10% on the actual headline) followed by Data Center's February 2026 hike (15% on standard, up to 40% on legacy). At least Atlassian is honest about the price step. Bundling is the silent version.
3. The new floor tier loses key features
This is the squeeze move. Plus on Notion still costs $10, but now Plus users have no AI access at all. Plus didn't get cheaper to compensate. The features users actually want (AI) moved up a tier. Plus is functionally a different product than it was 18 months ago.
Kit (formerly ConvertKit) ran a similar play with automations in 2024. Beehiiv moved custom domains and analytics behind paid the same year. In every case the floor tier still exists at the same price; it just does less of what people sign up to do.
What to actually do
Three rules of thumb if your stack lives on tools doing this:
- If you're grandfathered on a legacy add-on price: do not cancel. The minute you cancel, the rate evaporates. Pay through holidays, pay through pivots, pay through quiet quarters. The compounding savings are larger than the temporary cost of keeping a tool you're barely using.
- If you're evaluating a new tool's AI tier, model the path forward, not the price today. If you might want AI in six months, treat the AI-included price as your real subscription cost. Stripe-Atlas-style "cheap to start, expensive at scale" is now a dominant pattern.
- Audit annually for tier creep. Look at the bundling diff between when you signed up and today. If a feature you assumed was included is now gated to a higher tier, the gate either pulled you up (you upgraded) or pushed you out (you used the feature less). Either way, your effective per-seat cost moved.
The deeper read
Founder-tools pricing in 2026 is shifting from "subscription with add-ons" to "tier with bundle." The tier with bundle is structurally more profitable for vendors because it forces upgrade decisions instead of allowing opt-outs. It's also more confusing for buyers because the headline numbers stop telling the truth about what changed.
The countermove for solo founders is simple: stop reading just the price column. Read the feature column too. Compare the feature distribution today to your sign-up day. If features you depend on slid up a tier, you're paying more than you signed up for, even if the headline number is the same.
If you want a fast scan across your whole stack, paste your tools into Stack Audit. It flags the bundling-trap candidates and tells you which ones are about to compress further.