Pricing an AI product: why our pilot is free

Typelessity prices the pilot free with no time limit: the whole booking engine, up to 50 booking requests a month. It is not a feature-gated freemium tier and not an expiring free trial. The paid tiers are published — Starter €39, Pro €149 and Enterprise €399 a month — and a business moves up when its volume passes the cap. The reasoning: standard SaaS pricing assumes zero marginal cost, screenshot-proven value, and credit-card procurement. None of those hold cleanly for B2B AI in 2026, so value is proven on the customer's own site before any price decision. Read the pricing page for tier shape.

The conventional SaaS pricing playbook says: a free tier, a starter plan, a growth plan, then a plan on request. Typelessity keeps the ladder but changes the first rung: the pilot is free with the whole booking engine, no time limit and no credit card, and every paid tier is published. This article explains the reasoning. We have no conversion data from it yet, so it quotes none.

What does "free pilot" actually mean at Typelessity?

A Typelessity pilot is the entire product on a real customer-facing surface, set up by the business in the Webappski portal, with support by email. Not a sandbox. Not a demo. Real customers sending real booking requests — up to 50 a month — processed by Typelessity's infrastructure, at Typelessity's cost.

The pilot has no time limit. A business moves to a paid tier when its monthly booking requests pass 50 — after value has had time to show on its own data. The risk sits on the vendor side. That is the feature, not a side-effect.

Bottom line: the pilot is the entire product, not a watered-down preview. A pilot that hides features is a free trial pretending to be a partnership.

Why does this work for B2B AI specifically?

The historical SaaS pricing model assumes:

1. Marginal cost of serving one more user is approximately zero. Cloud infrastructure plus storage; pennies per user.
2. The product's value is obvious from a screenshot. A user can evaluate it in a minute.
3. Procurement decisions are made by individual users with credit cards. PLG fits.

None of this is true for AI products in B2B:

1. Marginal cost is real. Model inference fees per request: every booking request costs model fees.
2. Value depends on how the model performs on the customer's specific data. Staff names, service terminology, address formats, language mix — none of which can be screenshot-tested.
3. B2B AI procurement involves legal review, data residency questions, IT security questionnaires, and DPA negotiation. Credit-card PLG is dead in this segment.

So the customer has to be allowed to prove value before procurement. A free pilot is procurement-compatible — it slots into the existing evaluation framework. A subscription as the first step is not — it requires a procurement decision before value is proven. That is why Typelessity's paid tiers come after the pilot, not instead of it.

How does the unit economics math actually work?

Per-pilot marginal cost is dominated by model inference fees plus infrastructure. A pilot is capped at 50 booking requests a month, so that cost is small compared to the cost of a single sales call, let alone a full procurement cycle.

The real constraint is human attention per pilot, not compute. A pilot the business sets up itself scales — many can run in parallel. A pilot that needs hours of our engineering each does not — the team would be bankrupt before the first conversion.

The model fits Typelessity because:

• The widget is a single embed.
• The config is JSON, generated by the portal's setup wizard.
• The backend integration is a REST POST (JSON) to the customer's existing booking system, if it has an API, or an email to the business.

If the integration cost per customer were higher, the pilot model would not work.

Bottom line: the constraint is human time, not money. Free pilots scale only when integration is cheap.

What is the Enterprise tier?

Enterprise is published like every other tier: €399 a month for up to 6,000 booking requests, with unlimited sites. What a list price cannot fix is agreed per contract, on request:

• Volume above 6,000 booking requests a month.
• A named contact person.

An uptime commitment is not on that list. No availability SLA in the standard terms yet; one becomes possible once the hosting plans support it.

There is no on-premise or sovereign deployment: Typelessity is hosted only, with the service in Frankfurt and the database in Ireland (see GDPR and an AI booking widget). For tier shape, see Pricing.

Why "no time limit" specifically?

A 60-day free pilot reads as a discounted subscription, and the customer treats it as one.

A pilot with no time limit reads as a partnership. Two reasons:

1. Trust. "Free for 60 days" is a deadline imposed by the vendor. "Free pilot, pay when you are ready" is a commitment imposed by the customer. Different procurement category.
2. Fiscal-quarter alignment. When the deadline is removed, customers can convert when their internal budget cycle allows, not at an arbitrary calendar cutoff imposed by the vendor's marketing team.

Bottom line: the deadline is the friction. Removing it surfaces the customer's actual readiness signal.

What this is and is not

It is not feature-gated. Classic freemium keeps features behind the paid tier. Every Typelessity tier, the pilot included, runs the same booking engine; the tiers differ in booking requests a month and the number of websites. The pilot is not a lesser version.

It is not a free trial. A trial expires. The Typelessity pilot does not.

It is a volume ladder. The business pays once its volume passes the pilot's 50 booking requests a month. Until then the risk sits on the vendor. That is the structural feature.

Direct comparison summary

How we rank pricing models for B2B AI products by procurement compatibility (our view, not a study):

• Free pilot, no time limit, published tiers → highest procurement compatibility
• Annual Enterprise contract with proof-of-concept clause → strong; legal-friendly
• Time-boxed free trial (30-90 days) → medium; reads as discount, not partnership
• Freemium (feature-gated free tier) → weak for B2B AI; the gating creates the wrong evaluation signal
• Per-seat monthly subscription → fails procurement gate for AI products
• Pure usage-based (per-token / per-call) → works for self-serve developer tools, not for AI booking

For a self-serve developer tool, usage-based is the right call. For B2B AI booking, a free pilot in front of published tiers is the model we chose.

When the free-pilot model is the wrong choice

• Marginal cost per user is high (GPU-heavy video generation, frontier-model agents). Pilots eat the budget.
• Customer base is huge and pilot-quality support cannot scale. Self-serve metering is the right primitive.
• The product is naturally usage-priced (per-API-call, per-token). Publish a free tier and meter usage from day one.
• Sales cycle is short and credit-card-driven (consumer SaaS). A free pilot in front of the paid tiers adds friction with no gain.

For lower-touch products, publish a free tier and meter usage. For mid-market B2B AI with model inference at the core, a free pilot in front of published tiers fits the customer's procurement reality.

FAQ

Why does every Typelessity tier start with a free pilot? B2B AI procurement needs value proven on the customer's own data before a price decision. The pilot comes first; the published paid tiers come after it.

What does a "free pilot" actually include? The whole booking engine, on a real customer-facing site, up to 50 booking requests a month, set up by the business in the portal with support by email. No time limit.

How does Typelessity make the unit economics work on free pilots? A pilot is capped at 50 booking requests a month, so its cost is bounded. The real constraint is human attention per pilot, not compute; self-serve setup is the prerequisite.

What does Enterprise include, and what is agreed per contract? €399 a month for up to 6,000 booking requests and unlimited sites. Volume above that and a named contact are agreed per contract, on request. No availability SLA in the standard terms yet; one becomes possible once the hosting plans support it.

When is the free-pilot model the wrong choice? High marginal cost per user, huge customer base, or natively usage-priced products. Publish a metered free tier instead.

For the architectural reason single-call extraction makes pilots cheap, see Replacing a booking form with one GPT call. For what to check on data protection, see GDPR and an AI booking widget. For tier shape, see Pricing.

— Alex Isa, founder of Typelessity. Also founder of Webappski and TypelessForm.