By Saga Lindqvist, CMO at Stomme AI
Every SaaS company has a free tier. It's in the playbook: give something away, build a user base, convert a percentage to paid. Freemium is the default growth model.
We don't do this. And we won't.
Not because we don't understand freemium economics. Because a free AI agent is fundamentally different from a free project management tool or a free email client.
A free agent is a broken promise
A free version of Notion still works. It has fewer features, but the core product — writing documents, organising notes — functions perfectly. The paid version adds convenience and collaboration. The free version delivers genuine value.
A free AI agent doesn't work that way. An agent needs to run continuously. It needs to maintain context over days and weeks. It needs to connect to your email, your calendar, your files. It needs persistent memory and background processing.
Strip any of those out, and you don't have a "limited but useful agent." You have a chatbot with extra steps. And we already have plenty of those.
A degraded agent is worse than no agent at all — because it creates the expectation of autonomous operation and then fails to deliver it. The user thinks "agents don't work" when the real problem is "this agent isn't resourced to work."
The compute reality
Here's the math most freemium AI companies don't talk about.
Running an AI agent costs real money. Every interaction with a language model incurs API charges. A Personal tier agent on a moderate-use day might process 50-100 substantial requests. At current API rates, that's €2-5 per day in compute costs alone — before infrastructure, support, or maintenance.
At €45.99/month, we operate on margins that are viable but not generous. A free tier wouldn't be "limited" — it would be subsidised by paying customers. And we don't think it's fair to make paying customers fund a service for people who haven't committed.
Quality requires commitment — from both sides
Every Stomme AI customer gets personal onboarding. A 45-minute call where we configure your agent to match your workflow, your communication style, your tools. That's an hour of our team's time per customer.
We can't do that at scale for free users. And skipping it — letting users self-configure — produces worse outcomes. An agent configured with context beats an agent configured by defaults, every time.
The onboarding isn't overhead. It's the product. And products that matter cost money.
What we chose instead
Instead of a free tier with degraded quality, we chose:
Founding pricing. 50% off your first month. Low enough to try without risk. High enough to ensure you're committed.
No lock-in. Cancel any time. If the agent doesn't deliver value, you stop paying. No annual contracts. No cancellation fees.
Transparency. We publish what you get at each tier — capacity, capabilities, support level. No hidden limits, no surprise usage caps, no "contact sales" walls.
The alternative to free isn't expensive. It's honest pricing for a real product.
The freemium graveyard
Look at your app drawer. How many free AI tools have you signed up for and forgotten? How many are "free tier" accounts you'll never upgrade because the limited version was underwhelming enough to abandon?
That's the freemium trap. Free users don't convert because the free product doesn't deliver enough value to create demand for more. But it delivers just enough to inflate signup metrics.
We'd rather have 100 paying customers who use their agent every day than 10,000 free signups that churn in a week. The number that matters isn't signups. It's daily active usage.
The confidence test
Here's how we think about it: if your product is good enough to charge for, charge for it. If it's not good enough to charge for, a free tier won't fix it.
We charge because we believe the product is worth it. And the customers who pay confirm it — not because they're locked in, but because the agent genuinely delivers.