TrustMint

AI Transparency

Last updated: October 8, 2026

AI Services We Use

TrustMint uses OpenAI for Google review drafts, Google Gemini for private-feedback replies and email templates, and TypeSafe Jev through Cloudflare for decisions.

What AI Is Used For

We use OpenAI to:

  • Draft Google review reply suggestions when you request a draft.

We use Google Gemini to:

  • Generate email template suggestions for review request campaigns.
  • Draft reply suggestions for private customer feedback.

We use Jev through Cloudflare to:

  • Classify customer feedback into themes, sentiment, priority, requested action, recovery potential, and possible safety, legal, spam, or unresolved issues when the account owner enables Voice of Customer insights.
  • Check private-feedback reply drafts for relevance, unsupported claims, tone, privacy, language, and other safety concerns before the draft is shown. Google review drafts rely on merchant review without a separate AI verification request.
  • Map an unfamiliar Google Business Profile category to TrustMint's business categories when Google's stable category identifier is not already recognized.

Content generated by OpenAI or Gemini is presented as a suggestion that you can edit, accept, or discard before it is sent. Jev returns predefined labels and probabilities; it does not generate customer-facing content, publish anything, or take actions on its own.

What Data Is Shared with AI

We send only the context needed for the AI feature you choose to use:

  • Email-template generation uses business details and the instructions you provide; it does not use a customer's email address or phone number.
  • Reply drafting uses the business details, rating, and review or private-feedback text so the suggestion can address the specific experience. TrustMint replaces known names, email addresses, and phone numbers with placeholders before sending Google review drafts to OpenAI or private-feedback drafts to Gemini, then restores the first name locally in the generated draft.
  • Reply verification sends the minimized source context and proposed draft to Jev through Cloudflare after the same identifier replacement. Jev returns a predefined route and quality or safety probabilities; it does not rewrite or publish the reply.
  • Voice of Customer analysis sends the review or private-feedback text, rating, source, business category, and configured language to Jev through Cloudflare. TrustMint removes known names, email addresses, and phone numbers first. Jev returns only predefined labels and probabilities; it does not generate customer-facing text.
  • Google category mapping sends only the public business name and the primary and additional categories supplied by Google to Jev through Cloudflare. This fallback runs only when TrustMint cannot map Google's stable category identifier directly.

AI drafting is optional and initiated by the account user. Voice of Customer analysis runs only after the account owner enables it and can be disabled at any time. Do not submit health information, payment details, or other sensitive personal data for AI processing. The AI integration and Google Contacts import are separate features; contact lists are not shared with these AI services.

Google review analysis and AI reply drafting are available for eligible accounts. Drafts and individual assessments derived from Google reviews are deleted with the source review and no later than the 30-day Google review-data window. Content-free aggregate metrics and trends may be retained for historical reporting. You review every AI draft before publishing it.

Data Processing

OpenAI processes Google review drafting requests through its Responses API with output storage disabled. This setting does not change OpenAI's separate abuse-monitoring retention policy. See OpenAI's data controls. Gemini processes private-feedback drafting and template requests under Google's API Terms of Service. TypeSafe processes decision requests using Jev through Cloudflare AI Gateway with Cloudflare-managed provider credentials. Our requests disable Gateway request logging, payload logging, and response caching, and request no-retention routing. Cloudflare lists Jev as supporting zero data retention for this managed route. TypeSafe states that it does not train Jev on customer requests or responses. Processing may take place internationally under the applicable Cloudflare Data Processing Addendum and TypeSafe data-processing terms. See Cloudflare's managed-provider retention rules. We use AI outputs only to provide the enabled TrustMint features.