Legal · AI

AI Disclosure & Acceptable Use

This disclosure explains how the Thorbis AI assistant (“Brain”) works: the autonomy modes you choose, the guardrails that limit what it can do on its own, why AI output must be reviewed before you rely on it, and your responsibility for every AI-assisted message and action sent under your account.

Effective June 23, 2026Updated June 23, 2026v1.0

This document is provided for transparency and does not constitute legal advice. It describes how Thorbis operates; your use of the platform is governed by the agreement you accept.

1. What the AI features are

The Service includes AI features, including an assistant referred to as “Brain.” These features are powered by third-party large language models — currently models such as Anthropic's Claude, accessed through Vercel's AI Gateway. The underlying models and providers may change over time, and Thorbis does not guarantee that any particular model or provider will be used.

Depending on how you configure the Service, the AI features can draft and send messages, summarize information, recall context from your account, and suggest or take certain actions within your account. The specific tools available, and what each can do, depend on your configuration and the plan and features you have enabled.

2. You control how much the AI does on its own

You configure how much the AI may do on its own. Autonomy is a setting you choose — generally on a per-feature basis — and the Service offers a range of modes from fully passive to automatic:

Observe-only — the AI watches and gathers context but never acts on its own.
Suggest — the AI proposes actions for a person to review and take; nothing happens until a person acts.
Approval-required — the AI drafts a message or action, and a person must approve each one before it happens.
Limited automatic — the AI automatically executes only certain low-risk actions, within configured limits.
Full automatic — the AI automatically executes actions within guardrails; this carries higher risk and requires your explicit opt-in.

These are configurable settings that you, as the Customer, choose. The names and exact behavior of each mode may evolve as the Service changes. When you select a more automatic mode, you accept the corresponding increase in risk that an action may be taken without a person reviewing it first.

3. Guardrails that apply in every mode

Regardless of the autonomy mode you select, certain hard limits apply to AI-initiated activity:

The AI will not contact people you have marked as do-not-contact.
Automated customer-facing sends are capped per time window to limit the volume the AI can send on its own.
Money-affecting actions are subject to a spending ceiling.

Guardrails reduce risk — they do not eliminate it

These guardrails are designed to limit the impact of an automated mistake, not to guarantee that one never happens. You remain responsible for reviewing AI activity and for configuring autonomy in a way that is appropriate for your business.

4. AI output is provided “as is”

AI output may be wrong — review before you rely on it

AI output is generated by probabilistic models and may be inaccurate, incomplete, outdated, biased, or inappropriate. It is provided without warranty and is NOT legal, financial, tax, accounting, medical, or other professional advice. You are responsible for reviewing AI output and for any message sent or action taken — including by automatic modes you enable — as if you had done it yourself.

Because output is generated by predictive models, the same prompt can produce different results, and results may reflect errors or limitations in the model. Do not treat AI output as a statement of fact verified by us.

5. Your communications, your responsibility and consent

AI-assisted messages are your communications. They are sent under your account and on your behalf. You remain responsible for having the consent required for each communication and for compliance with all applicable laws — including, without limitation, the Telephone Consumer Protection Act (TCPA), CAN-SPAM, and similar communications and consumer-protection laws — for anything the AI sends on your behalf.

The AI features, and your use of them, are subject to the Acceptable Use Policy. You may not use AI features to do anything that policy prohibits, and enabling a more automatic mode does not relieve you of responsibility for the resulting communications or actions.

6. How data is handled

To generate responses, your prompts and relevant context are sent to AI subprocessors that provide the underlying models and gateway. Customer Data processed by AI features is handled under our Privacy Policy and Data Processing Addendum, and is shared with AI subprocessors only to the extent needed to provide the feature.

We do not sell your data. Processing of prompts and context by an AI subprocessor is governed by that subprocessor's applicable terms. The current AI subprocessors we use are listed in our Subprocessors page, and our handling of personal data is described in our Privacy Policy.

See the Subprocessors page (/legal/subprocessors) for the providers that support AI features, and the Privacy Policy (/legal/privacy) for how we handle personal data.

7. Not a substitute for professional judgment

AI features are tools to assist you, not a replacement for your own judgment or for qualified professional advice. Do not rely on AI features for high-stakes legal, financial, safety, or compliance decisions without independent human review. Where a decision carries legal, financial, safety, or regulatory consequences, a qualified person should review the output before you act on it.

8. Availability and changes

AI features and specific tools may change, may be limited, or may not yet be available, and they may be added, modified, or removed over time. Some AI features are metered and may incur usage charges in addition to your base plan, as described in the billing section of the Terms of Service. Where a feature is described as not yet available, it is offered subject to its actual release and may differ from any preview description.

9. Contact

Questions about the AI features or this disclosure: legal@thorbis.com. For help configuring autonomy modes or guardrails: support@thorbis.com.