The agreement
These terms are between [legal entity name], registered at [registered address] (“ControlClaw”, “we”) and the organisation that signs up (“you”). Creating an account, or using the service, means you accept them. If you accept on behalf of a company, you are saying you may bind it.
How we handle data is in the Privacy Policy, which is part of this agreement.
What the service is
ControlClaw provisions and manages machines that run an AI agent for you, and gives you a console to work with them: the agent itself, a firewall box that filters what your agents can reach, backups, and the integrations you switch on.
The agent software is a separate open-source project we host for you; we are not affiliated with it. We may change how the service works, and add or retire features. If we remove something you rely on we will tell you first.
Your account and your people
- Give accurate details and keep your sign-in secure. Tell us promptly if you think someone else has got in.
- An organisation has owners, admins and members, and those roles decide who can change what. You are responsible for who you invite and for what they do with the access you give them.
- You must be able to form a contract, and you must not be barred from using the service under applicable sanctions or export law.
Your content, and what your agents do
Everything you or your agents put on your machines stays yours. We claim no rights over it beyond what running the service requires, and we do not use it to train models.
An agent acts on your behalf. You decide what it may reach, which accounts it holds and which of your people can drive it. What it does with those permissions is your responsibility, including anything it sends, deletes or spends. Two practical consequences worth stating plainly: an email, a web page or a document an agent reads is untrusted input that may try to instruct it, and an account you connect is available to the agent and to everyone in your organisation who can open it.
You are responsible for having the right to connect the accounts and data you connect, and for telling the people whose data your agents handle, where the law requires it.
What you must not run on it
Do not use ControlClaw, or let an agent use it, to:
- break the law, or infringe somebody’s rights;
- attack, scan or overload anyone else’s systems, or send spam or bulk unsolicited messages;
- mine cryptocurrency, resell raw compute, or run workloads whose purpose is to consume resources;
- circumvent your own firewall rules, the model allow-list, or the limits of your plan;
- attack our systems or another customer’s. Testing the security of your own boxes is fine — tell us first, and stay inside them;
- store or process data you are not allowed to put on infrastructure like ours.
If something you run threatens the platform or another customer, we may suspend it without waiting. We will tell you why, and we will put it back as soon as it is safe.
Plans, payment and cancellation
- Plans and prices are on the pricing section. Subscriptions are billed in advance through Stripe and renew automatically until you cancel.
- A plan includes an allowance of AI tokens and a number of agents. Going past the allowance is handled by your firewall — agents with their own provider key keep working — and extra agents need the larger plan.
- A trial, where you are given one, runs for 14 days and needs no card. Add a payment method before it ends and the plan carries on. If you do not, the subscription cancels rather than charging you.
- Cancel at any time from your billing settings. Cancelling takes effect at the end of the period you have paid for; your boxes and their data go with it, so take a backup first if you want one.
- Prices can change. We will give you [notice period] before a change affects a renewal.
- Fees are exclusive of tax, which is added where it applies. Payments are [refund policy].
- If a payment fails and stays unpaid, we may suspend the service and, after that, delete the machines.
Third-party services
When you connect Google, a model provider, a messaging channel or an app, you are using that provider under its own terms and within the permissions you granted it. We pass the credential to your firewall and get out of the way; we do not control what the provider does, what it charges, or whether it stays available. A provider revoking access, changing its API or refusing a request is not a failure of this agreement.
Availability, backups and support
We work to keep the service up and do not promise a particular uptime. [There is no separate SLA yet.] Maintenance, provider outages and your own configuration can all interrupt it.
Backups run daily, before an in-place update, and whenever you ask. They are encrypted on your box with a key we do not hold, which is the point — and it means that if you lose your recovery key, we cannot restore for you. Keep it somewhere safe.
Support is through the console chat and [support contact email], during [support hours].
Suspension and ending it
You can stop at any time by cancelling. We can suspend or end an account that breaks these terms, does not pay, or puts the platform or another customer at risk — with notice where we reasonably can, and immediately where we cannot.
When the agreement ends we delete your machines and the data on them. Ask us before you go if you need an export; afterwards there may be nothing left to give you.
What we do not promise
The service is provided as it is. To the extent the law allows, we make no implied warranties of merchantability, fitness for a particular purpose or non-infringement.
AI output is not reliable by default. An agent can be wrong, can be misled by content it reads, and can take an action you did not intend. The approval and firewall features exist to bound that, and they reduce the risk rather than removing it. Do not put an agent in charge of something where a mistake is not recoverable, and do not treat its output as legal, financial, medical or safety advice.
Liability
To the extent the law allows, neither side is liable for indirect, incidental or consequential loss, or for lost profits, revenue or data, and each side’s total liability under this agreement is capped at [liability cap]. Nothing here limits liability that cannot be limited — including for death or personal injury caused by negligence, or for fraud. You will cover us for third-party claims arising from your content, your agents’ actions or your breach of these terms.
Law, disputes and changes
This agreement is governed by the law of [governing law], and disputes go to the courts of [jurisdiction]. Talk to us first — most things are a misunderstanding that a conversation fixes.
We may update these terms. The date at the top changes, and for a material change we will tell the owners of each organisation before it takes effect. Carrying on using the service after that means you accept the new version.
If a clause turns out to be unenforceable, the rest still stands. You may not transfer this agreement without our consent; we may transfer it as part of a merger or sale of the business.
Questions about these terms: [legal contact email].