One bill for AI across the whole company

A credit is appto’s unit of work — you spend one when the AI does something for you: writes an email, runs an analysis, searches your documents. A reply to a customer email is about a hundred credits, a report off a sales spreadsheet a few hundred, an audit of sixty product listings a few thousand. One rate, in the open: 20 PLN per 10k credits.

There is no single number here, and nobody honest will give you one. The same request costs differently on a light model and on a strong one, and it also matters how much context appto has to read: a short email is a different bill entirely from an analysis of a hundred-page report. So instead of promising a task count, we give you 10,000 credits to start — you get to see your own usage before you pick a tier. And if you still get it wrong, extra credits are at the same rate as the ones in the pack.

Not the way a typical SaaS does it. The platform fee is one fee for the whole team and follows the tier you are on — not how many people actually sign in. Within a tier, adding one more person does not change the bill, and credits only come off for tasks you hand over. So you can give access to the whole company, not only to the people you are sure will use it.

Yes — and it is the one place in this pricing where the differences are large. appto runs models across five price tiers: the same task costs about fifteen times more on the strongest than on the cheapest. All of them (Claude, GPT, Gemini, Kimi) are yours at no surcharge and you switch between them in the conversation bar, so the bill comes down to how often you have to reach higher. Beyond the model, it also matters how much context appto has to read and how long the process runs.

It would, if the model were handed the task the way an ordinary chat hands it over. In appto it does not work alone: it sits inside our own harness — a layer that walks it through the job and makes part of the decisions for it. That layer breaks the request into steps, picks which tool to reach for and pulls the data at source — from the CRM, the mailbox, the drive — so the model neither hunts for it nor guesses. From the knowledge base it gets the three paragraphs that matter instead of a hundred-page report, and from the assistant the goal, the rules and the answer format. None of that makes the cheaper model smarter — it takes off it the part of the work that used to force you up a tier. What is left is a narrowly framed job, and a model from the bottom of the ladder handles that; which is why the cheapest one is appto’s default. Some work really does need the strongest model — then you switch to it in one click, at no surcharge.

Work does not stop. You buy another tranche mid-month at the same rate as in the pack — 20 PLN per 10k — or move up a tier from the next period. You decide when and how much.

No — each period starts on a full limit. That is why it is better to size the tier to your real usage and top up when needed than to buy a stockpile up front. You can change the tier at any point, effective from the next period.

No. Every tier is the full appto: all models, context memory, integrations with your tools, assistants per role, multi-step automations, SSO login, permissions and audit log, EU servers. Tiers differ only in the number of people and the size of the pool.

The plans in the configurator cover teams up to about 100 people. For a larger organization or a dedicated rollout — SLA, your own integrations and models, priority support — we quote individually. Get in touch and we will put together an offer for you.