Pricing
The jobs your team repeats every week - the Monday report, the overdue invoices nobody gets round to chasing, the support mail that arrives overnight - handed to agents that run them on a schedule, the moment a message or a web request arrives, or at the next check of the mailbox or the folder you point them at. One workspace for the whole team, and no seats to count.
Free
- One workspace.
- Memory and storage are capped, at a size meant for trying it properly.
- The two models a workspace needs: one for everyday work, one that also reads pictures and listens to recordings.
- An allowance of model use included, with more by arrangement.
Pro
- As many workspaces as you like.
- More memory and storage, for a team using it every day.
- Every model our provider carries, rather than a chosen pair.
- Models from AnthropicOpenAIGooglexAIMetaMistralDeepSeek For exampleClaude Fable 5.1Claude Opus 5GPT-6 AstraGemini 3.8 FlashGrok 4.6Llama 4 MaverickMistral Medium 3.5DeepSeek V4.1 Flashand every other model our provider carries.
- A larger allowance of model use included, with more by arrangement.
Enterprise
- As many workspaces as you like.
- Memory and storage sized with you.
- Any model, including your own account with a provider if you would rather use one.
- Model use sized with you.
- Hosted by us, or on your own infrastructure if a policy or a jurisdiction requires it.
These are the prices you pay. We are not registered for VAT, so none is added to them, and nothing else is charged on top.
In every plan
Every plan is the whole product. What changes between them is how much room you have, how much model use is included and which models you can reach - not which parts are switched on.
You get a private workspace at a web address you choose. Your team signs in there, and it is the same address your webhooks and your own machines talk to. Your files, your data and your team's work stay separate, and are never mixed in with another customer's.
The AI comes with it: no separate ChatGPT or Claude subscription to hold, no API key to manage and no second invoice to reconcile. That is the default rather than the only way - if you would rather bring your own account with a provider, or need a particular model we do not carry, say so and we will arrange it.
Hand over as many routines as you like: the Monday report written before you get in, the overnight support mail triaged, the overdue invoices chased. Each one runs on a schedule, the moment a message or a webhook arrives, or at the next check of a mailbox or a watched folder.
The ceilings are yours to set - on the month's AI spend, described below, and, if you want them, on what a chain of delegated work or a single workflow run may spend before it stops.
Agents learn your team's way of doing things. Not only the documents you hand over, but what was settled in a conversation in March, still informing the work in September.
Each agent gets a computer of its own to work on, so it runs your scripts, installs what it needs and hands back the finished result rather than a description of one. Point it at a machine you already run instead - a build box, a server in your office - and it works there. It dials out, so there is no inbound port to open and nothing to expose to the internet.
Take a copy any day you like. Your agents, your workflows, your scheduled tasks and everything your memory has built up come out in one file from Settings, and it is yours to keep.
Not sure it fits? Tell us the routine that eats your week and we will say whether this takes it off you.
What one of these looks like
You say it once, in a sentence: "On the first of each month, find invoices more than thirty days overdue, check whether payment has since arrived, and email a polite reminder only to the ones still outstanding." An agent builds the routine, you look it over, and from then on it happens on the first of every month whether or not anybody remembers.
That is one routine. The price is the same whether your team runs one or fifty.
The AI, and what it cannot cost you
You do not need an account with an AI company, a card with one, or an API key. Your workspace comes connected and the cost is part of what you pay here.
The models are picked for the two jobs a workspace needs: answering quickly, and working from images and voice notes. The current list is shown in your workspace wherever a model is picked, and explained on the Models page in the documentation.
A workspace can be given a monthly ceiling on AI spend, and it is off until an administrator sets one in Settings. The same screen shows what has been used so far this month, and an administrator can change the figure whenever you want a tighter one. The administrator also chooses what reaching it should do: refuse new runs until the month turns, or let the work carry on and send a warning instead. Nothing you have already done is affected either way.
Need a particular model, your own account with an AI provider, or the whole workspace running on your own infrastructure? All three are arranged for larger teams. Tell us what you need and we will set it up.
Nothing is locked in
Cancelling leaves your workspace running to the end of the month you have paid for. Nothing renews quietly, and nothing is charged after you cancel.
Take the export first: your documents, your agents and what they have learned all come out in one file, and it is yours to keep. Ask for the conversations and the machines you attached and they come with it.