Contract terms

1. What this page is

This page sets out the standard commercial terms GuardBotAI proposes in enterprise agreements. It is a plain-English summary provided so you can review our position before procurement, not a contract in itself.

Nothing on this page creates an agreement. Coverage, obligations, and remedies exist only under a written agreement executed by both parties, and that executed agreement prevails over anything stated here or elsewhere on this website.

2. Scope of the service

GuardBotAI is a hosted platform. Customers connect their own AI applications and agents to the GuardBotAI gateway. Requests sent through the gateway are evaluated against the policy configured for that project, and the platform allows, flags, redacts, or blocks them according to that policy and the project's enforcement mode.

GuardBotAI only evaluates traffic that is actually routed through the gateway. Requests your systems send directly to an AI provider, bypassing the gateway, are outside the service entirely. Keeping your applications pointed at the gateway is the customer's responsibility.

3. Both software and AI make mistakes

GuardBotAI is human-made software that uses artificial intelligence. Human-built software makes mistakes, and AI systems make mistakes. No security control detects every attack, and no control avoids every false result.

We endeavour to build the very best software and AI available to us, and we test what we ship. We do not represent that GuardBotAI will detect all prompt injection, prevent all data exposure, or stop all misuse of an AI agent. Any statement to the contrary, wherever it appears, is not a term of the agreement.

4. Availability and response

Enterprise agreements may include availability targets and priority response targets. Those targets are stated in the executed agreement with their measurement method, exclusions, and remedies.

Planned maintenance, customer configuration errors, customer network faults, and failures of upstream AI providers are excluded from availability measurement.

5. Warranty and indemnity

Where an enterprise agreement includes warranty or indemnity coverage, that coverage applies only under the executed agreement and only where the customer meets its eligibility, deployment, configuration, and notification requirements.

Typical requirements include: keeping affected projects in enforcing mode, routing the affected traffic through the gateway, applying the policy baseline agreed at onboarding, maintaining a current subscription, and notifying us in writing within the notice period stated in the agreement.

Coverage does not apply to loss arising from traffic that never passed through GuardBotAI, from disabled or weakened policies, from containment being switched off, from misuse of the platform, or from a customer's own systems, models, or third-party providers.

6. Limitation of liability

To the maximum extent permitted by law, GuardBotAI is not liable for indirect, incidental, special, consequential, or punitive loss, nor for loss of profit, revenue, goodwill, or data.

Where liability cannot be excluded, our aggregate liability is capped at the amount stated in the executed agreement, which is ordinarily the fees paid by the customer in the twelve months before the claim.

Nothing in the agreement limits liability that cannot lawfully be limited, including liability for fraud, or rights a customer has under consumer law that cannot be excluded.

7. Making a claim

A claim must be made in writing to our team, within the notice period stated in the executed agreement, and must include the affected project, the time range, and the security event records concerned.

We will review the tamper-evident event records for the period in question and respond in writing with our findings. Where records show the traffic did not pass through the gateway, or that enforcement was disabled, we will say so plainly and explain the basis.

8. Customer data

We do not sell customer data and we do not train AI models on customer data.

Security event records store a hash of the request and, where a policy requires it, a redacted excerpt. Raw prompts, credentials, and secrets are not stored in event records. Retention periods follow the plan or the executed agreement.

Where the endpoint sensor add-on is used, the sensor reports findings only: the finding type, severity, location, and a one-way fingerprint. It does not send file contents, prompts, model responses, or credentials.

9. Fees, term, and termination

Fees, billing interval, usage allowances, and any overage treatment are stated in the executed agreement. Annual enterprise terms are invoiced in advance unless agreed otherwise.

Either party may terminate for material breach that is not remedied within the cure period stated in the agreement. On termination, access to the hosted platform ends and the customer may request an export of its security event records within the period stated in the agreement.

10. Governing law and contact

The governing law and jurisdiction are stated in the executed agreement.

GuardBotAI is an enterprise security product by Venture Group AI. For contract questions, contact sales@guardbotai.com or use the enterprise quote form.