Most SaaS pricing is simple. You pay per seat, per month. Maybe there is a usage component for storage or API calls. The model is well understood and investors like it because it scales linearly with headcount.
AI products break this model. A user who sends three messages a day and a user who runs twenty campaigns through the system are not the same customer. Charging them the same price means you either leave money on the table or price out the light user. Per-API-call billing solves the fairness problem but creates a worse one: the user starts thinking about cost every time they interact with the product. That is the opposite of what you want from an AI assistant.
Cleo uses energy credits. Here is why.
The Problem With Seats
Cleo is a marketing operating system. One person at a small business might use it to manage their entire marketing stack: strategy, content, email campaigns, ad management, analytics. Another person might use it once a week to check performance numbers.
Seat-based pricing ignores this completely. Worse, it creates a perverse incentive at the product level. If revenue scales with headcount, the product should encourage teams. But Cleo's value proposition is the opposite. She replaces the need for a team. She is the marketing department. Charging per seat penalises the exact outcome the product delivers.
The Problem With Pay-Per-Call
The alternative is metered billing. Charge for what you use. This is honest, but it introduces anxiety.
I have used AI products with visible token counters. You start self-censoring. You shorten your prompts. You hesitate before asking a follow-up question. The meter running in the background changes the relationship between the user and the product from collaborative to transactional.
For a marketing director, that is fatal. The whole point is that you can think out loud, explore ideas, ask "what if" questions, and let the AI do work you would not have thought to ask for. A cost-per-interaction model punishes curiosity.
Energy as an Abstraction
Cleo's energy system sits between these extremes. Each plan comes with a monthly energy allocation. Different operations cost different amounts of energy. A simple question costs less than generating a full content strategy. Running an ad campaign analysis costs more than checking your calendar.
The user sees their energy balance but not the per-operation cost. They know roughly how much runway they have. They do not perform mental arithmetic before every interaction.
This creates the right behaviour. Light users stay on cheaper plans and never feel constrained. Heavy users naturally land on plans that match their usage. And nobody hesitates before talking to Cleo because they are worried about burning through credits on a question.
The Pricing Tiers
Four tiers. Free at 50 energy per month, enough to explore. Starter at $39 with 500. Pro at $99 with 2,000. Business at $249 with 10,000.
The energy costs are tuned so that a Starter user can comfortably run a small business's marketing. A Pro user can operate aggressively across multiple channels. Business is for teams that use Cleo as their primary marketing infrastructure.
The key decision was making the free tier genuinely usable. 50 energy is not a demo. It is enough to run onboarding, generate a strategy, create some content, and see real value. The conversion trigger is not hitting a wall. It is wanting to do more of something you already know works.
What I Learned
Energy-based pricing is harder to implement than it looks. You need to assign costs to every operation, and those costs need to feel fair even when the user can see them. An image generation that takes ten seconds should cost more than a text response that takes two. But a text response that required expensive reasoning and context retrieval might actually cost you more than the image. The user does not care about your costs. They care about perceived value.
I spent time calibrating the energy costs against user perception rather than compute cost. A full content strategy, the kind a consultant would charge thousands for, costs meaningful energy. A quick question costs almost nothing. This matches how users think about value, even when the underlying compute costs do not follow the same curve.
The other lesson is that energy creates a natural upgrade path. Users do not hit a feature wall where some capability is locked behind a higher tier. They hit a usage ceiling. The conversation is not "pay more to unlock ads management." It is "you are using Cleo a lot, and that is working. Here is more capacity." That is a much easier conversation.
The Honest Tradeoff
Energy pricing is less predictable for the business than seat-based pricing. Monthly revenue varies with usage patterns. Forecasting is harder. Investors used to ARR-per-seat metrics need education.
I decided that was an acceptable cost for a better product experience. The user should never think about billing when they are thinking about marketing. The pricing model should be invisible during the work and obvious when it is time to grow.
That is what energy gives you. A budget, not a meter.