AI Transparency

Trendy Data Visibility uses Anthropic's Claude to power the AI analytics feature available on our Flourish plan. Claude is a large language model built by Anthropic, a leading AI safety company.

When you ask a question through our AI chat feature, Claude analyses your business data and provides actionable insights in plain language. This page explains exactly how your data is handled throughout that process.

Our AI chat uses a multi-step process designed to protect your data at every stage:

1. You ask a question about your business performance through the chat interface.
2. We determine what data is needed to answer your question, querying only the relevant metrics from your connected platforms.
3. Your data is anonymised before being sent to Claude. All personally identifiable information (names, emails, order numbers, customer IDs, campaign names) is replaced with generic labels.
4. Claude analyses the anonymised data and generates business insights.
5. We restore the original identifiers in the response so it makes sense to you, then deliver the answer.

Before any data is sent to Claude, we apply a thorough anonymisation process. The following types of information are always replaced with anonymous labels:

Only non-identifying business metrics are shared with Claude — such as sales totals, revenue figures, ad spend, click rates, and dates. This means Claude never sees who your customers are, only the aggregated and anonymised performance data needed to answer your question.

Our system uses a strict allowlist that limits exactly which data fields can be retrieved. Only the following types of metrics are accessible to the AI:

• Shopify: Sales figures, order counts, discount amounts, shipping fees, refund totals, fulfilment statuses
• Meta Ads: Ad spend, impressions, clicks, reach, cost-per-click, conversions, ROAS
• Sage Accounting: Transaction totals, tax amounts, account categories, money in/out
• Google Ads: Campaign performance metrics, spend, clicks, conversions

Data access is further enforced at the database level through Row-Level Security (RLS), which ensures you can only ever access your own data — even the AI cannot query another user's information.

We use Anthropic's API to access Claude. Under Anthropic's commercial API terms, data sent through the API is not used to train or improve their models. Your business data remains yours.

Anthropic may retain API inputs and outputs for a limited period solely for trust and safety purposes (e.g. abuse prevention), as outlined in their usage policy. No anonymised data sent from our platform is used for model development.

• Authentication required: Only logged-in users with an active Flourish subscription can access the AI chat.
• Anonymisation before transmission: All personally identifiable information is stripped before data leaves our servers.
• Allowlist-based data access: Only pre-approved data fields can be queried — the AI cannot access arbitrary data.
• Row-Level Security: Database-level policies ensure users can only access their own data.
• Secure transmission: All communication with Anthropic's API is encrypted via HTTPS/TLS.
• No persistent storage of anonymisation maps: The mapping between real and anonymised identifiers exists only in memory during processing and is discarded immediately after.
• Usage limits: Monthly credit limits prevent excessive usage and help control costs.

The AI chat feature is entirely opt-in. It is only available on the Flourish plan, and you choose when to use it. If you do not interact with the chat, no data is sent to Claude. You are always in control of what questions you ask and which platforms you connect.

Claude is built by Anthropic. For more information on how Anthropic handles data, see the links below:

• Anthropic Homepage
• Claude Overview
• Anthropic Privacy Policy
• Anthropic Usage Policy
• Anthropic Commercial Terms

Questions?

If you have any questions about how we use AI or handle your data, please don't hesitate to get in touch.