Overview

Amazon has introduced a generative‑AI query builder for Amazon Marketing Cloud (AMC) that lets sellers craft sophisticated audience segments using everyday language instead of writing SQL code. Unveiled at CES 2025, the tool translates natural‑language prompts into complete database queries, shrinking tasks that once required hours of scripting into a matter of minutes. For sellers who rely on AMC for campaign insights, the change removes a major technical barrier and opens advanced analytics to a broader range of users.

Key Points

• Natural‑language audience creation — Sellers type a plain‑English description, and the AI automatically produces the matching SQL statement.
• Cross‑channel signal integration — Generated queries can combine streaming TV ad views, audio ad exposures, product‑page visits, and purchase history into a single segment.
• Step‑by‑step query explanation — The system displays how each clause was built, allowing sellers to validate logic before execution.
• Interactive guidance — Users may ask follow‑up questions about targeting logic and receive real‑time suggestions within the interface.
• One‑click deployment — Completed segments can be sent directly to Amazon DSP or the ads console for immediate campaign use.
• Editable output — Sellers retain full control to modify any AI‑generated SQL before it runs.

How the Query Builder Works

1. Intent capture — The seller enters a conversational request, such as “Create an audience of shoppers who viewed my product detail pages, saw my streaming TV ads, and have not purchased in the past 30 days.” The AI parses the intent, identifies key actions (page view, ad exposure, purchase lapse) and the relevant time window.

2. Data‑source mapping — Based on the parsed intent, the system selects the appropriate AMC tables (e.g., product_page_views, tv_ad_impressions, order_events) and determines the necessary joins. For the example above, it links page‑view records to ad‑impression records on the same user ID and filters out any order events within the last 30 days.