Overview

Amazon Marketing Cloud (AMC) has entered beta with an AI‑driven chatbot that translates everyday English requests into SQL queries for audience creation and product support. The feature is designed for sellers and advertisers who lack deep data‑science skills but want to leverage AMC’s measurement and segmentation power. Launching now, the tool promises faster insights without writing code.

Key Points

• Natural‑language audience builder — Users can type a phrase such as “customers who added my product to cart but didn’t buy in the past 14 days,” and the chatbot instantly produces a ready‑to‑run SQL statement.
• Live product assistance — The same chat window answers questions about AMC capabilities, from basic terminology to advanced attribution methods, pulling directly from Amazon’s documentation.
• Editable query output — After generation, the full SQL script is displayed, allowing beginners to accept it as‑is and experts to tweak joins, filters, or aggregation logic.
• Context‑aware recommendations — While a seller drafts a query, the assistant suggests complementary audience strategies, such as adding a look‑alike filter or recommending a measurement metric.
• Strict data isolation — The AI can only query the advertiser’s own AMC dataset; it never accesses another seller’s information, preserving the existing security model.
• Beta‑stage limitations — Complex multi‑step queries may need several attempts, and vague prompts can produce incomplete results, so users should validate outputs before large‑scale deployment.

How the AI Audience Builder Works

1. Input interpretation — The seller types a plain‑English request (e.g., “show me shoppers who viewed my summer dress but didn’t purchase in the last 30 days”). The model maps the request to AMC’s schema, identifying tables such as product_views and orders.
2. Query construction — Using a large language model trained on AMC’s query patterns, the system assembles the appropriate , , and clauses, automatically adding date filters and product identifiers. The generated SQL is displayed in the chat pane.