Answer
Who is liable if the AI makes a mistake?
AlwaysOnIT is responsible for the platform behaving as documented and for the agent staying inside its published scope and approval rules. You remain responsible for your own tenant, the permissions you grant, and the actions you approve. Liability is capped at the fees paid, as set out in the terms of service.
The design limits the damage before liability matters
The honest starting point is that no support arrangement — human or AI — makes mistakes impossible, so the useful question is how much a mistake can cost. Zero standing access, single-purpose tokens, and mandatory admin approval for destructive or bulk actions mean the worst plausible unilateral error is small and reversible. Anything with a wide blast radius needed your explicit approval first.
Where our responsibility sits
We are responsible for the security and correct operation of the support platform, for keeping the agent inside its published scope, for the encryption, logging and retention described in the trust centre, and for telling you promptly if an incident affects your data. If the platform does something it was documented not to do, that is on us.
Where yours sits
Your tenant configuration, backups, licensing and user hygiene remain yours, as does the decision about which permissions to grant and to which of your admins. When the agent asks a named admin to approve a destructive action, that approval is a decision made in your organisation. Nothing in an AI service transfers responsibility for your environment away from you.
Contractual position, in plain terms
Our terms of service cap liability at the fees paid in the preceding period, exclude indirect and consequential loss, and are governed by English law — standard for a service at this price point, and stated plainly rather than hidden. If your risk profile requires a higher cap, professional indemnity backing, or a negotiated SLA with penalties, an AI-only service at £4 per user is the wrong product and we would rather tell you that now.
What we do when something goes wrong
You get the full audit trail of what the agent did and when, without having to ask twice, and any incident affecting your data is reported to you promptly with the timeline. That transparency is the practical remedy that matters most in the first hour — long before any contractual argument becomes relevant.
Related
Follow-on questions
Do you carry professional indemnity insurance?
Ask us before you sign if insurance backing is a procurement requirement, and we will tell you exactly what is and is not in place rather than implying cover we do not hold. Our published liability position is the cap stated in the terms of service.
Is there an SLA with penalties?
The published service commitments describe availability and response expectations, but the standard subscription does not carry financial penalties for missed targets. There is also no contract term, so the practical remedy for poor service is cancelling at the end of the month.
Get started
Open a chat. The AI picks up before the second ring — every time.
No onboarding calls. No sales process. Connect your devices and start raising tickets in under ten minutes.