← Back to home

Responsible AI Policy

Last updated: July 26, 2026

Flow, provided by Jade’s Gems & SOULutions LLC, doing business as DivineX, includes optional AI-powered features. This page explains what they are, how they work, and their limitations — in plain language, matched to what’s actually built today. This policy is governed by the laws of the State of Texas, United States, on the same terms as Section 14 of our Terms of Service.

1. What Flow’s AI features do

All of Flow’s AI features share one persona per workspace (a system prompt you write, plus optional business hours, escalation rules, and a knowledge base drawn from your own website) — the same identity answers across whichever channels you enable:

2. How these features work, technically

Text-based AI replies (assistant, web chat, SMS, WhatsApp) are generated by sending the conversation and your configured persona to OpenRouter, a third-party AI model gateway, which routes the request to an underlying language model (Claude by default; you or we may configure a different model). Voice calls are additionally handled by Vapi, a third-party voice AI provider, which converts speech to text, generates a response, and converts it back to speech in real time.

3. Human oversight built into the product

Some AI actions require your explicit confirmation before they take effect, rather than executing automatically — this is a deliberate design choice for actions with real-world consequences (for example, creating records or sending messages on your behalf in certain flows). Outbound voice calling additionally passes through a compliance gate that checks consent, calling windows, and contact frequency before any call is placed — you remain responsible for your own compliance obligations (Section 5 of our Terms of Service) on top of this safeguard.

4. Limitations

AI-generated content can be wrong. It may misunderstand context, generate inaccurate information, or produce a response that doesn’t match your intent — especially for nuanced, sensitive, or high-stakes conversations. You are responsible for:

5. Data use by our AI providers

Conversation content you send through Flow’s AI features is transmitted to OpenRouter (text) and/or Vapi (voice) to generate responses, and is stored in our own database as described in our Privacy Policy. Text and voice are handled differently by these providers, based on each provider’s own published policy:

Text (OpenRouter). OpenRouter’s documented default is that it does not store submitted prompts or completions, and does not use them to train or improve models, unless data collection is explicitly enabled at the account level. Our integration does not enable that data collection, so text conversations fall under OpenRouter’s default (no storage, no training use) as a matter of configuration.

Voice (Vapi). Vapi’s published privacy policy states that call recordings, transcripts, and logs may be retained and used to help improve their AI services, unless a privacy-enhanced configuration is enabled for the account. We are evaluating enabling that configuration; until we do, voice call content may be used by Vapi consistent with their standard policy.

6. Model changes

The specific AI model or provider we use may change over time as we adopt better or more cost-effective options. Material changes to which provider(s) process your data will be reflected in this policy and our Privacy Policy.

7. Related Documents

This policy should be read together with our Terms of Service (Section 6) and Privacy Policy.

8. Contact

Questions about how Flow’s AI features work — email hello@divinex.io, or write to us at:

Jade’s Gems & SOULutions LLC
440 Louisiana St
Houston, TX 77002
United States