Meta Tests AI Shopping Assistant and Research Tool, Launches U.S. Experiments Today

Meta’s AI shopping assistant, dubbed “Meta Shop,” will appear as a chat‑style overlay on the company’s family of apps, letting users ask natural‑language questions about products while they browse. In the first U.S. experiment, the tool pulls data from merchants’ product catalogs and the user’s own activity to generate concise recommendations, price comparisons and availability alerts, according to a PYMNTS.com report. The assistant is also designed to surface related items and accessories, effectively turning a passive scrolling session into an interactive research experience.

The rollout is limited to a handful of participants in the United States, with Meta monitoring engagement metrics such as query length, click‑through rates, and conversion signals. Dataconomy notes that the company is testing the feature across both its core Facebook platform and the newer Threads app, aiming to gauge how the assistant performs in distinct social contexts. Meta’s internal teams are also using the same underlying model as a “research tool” for advertisers, allowing brands to query audience insights and content performance data without leaving the interface.

Behind the scenes, the assistant relies on Meta’s Llama‑2 family of large language models, fine‑tuned on e‑commerce datasets to understand product attributes, pricing structures and user intent. The company has not disclosed the exact size of the model or the volume of training data, but the PYMNTS.com article confirms that the system can retrieve real‑time inventory information from partner retailers via API integrations. This capability enables the assistant to flag stock shortages or price drops instantly, a feature that could reduce the friction that typically leads shoppers to abandon carts.

Meta frames the experiment as a step toward a broader “AI‑first” commerce strategy, positioning the assistant as a bridge between social discovery and purchase intent. By embedding the tool directly into its social feeds, Meta hopes to keep users on its platforms longer while capturing a larger share of the transaction funnel. The company has not announced a timeline for a public launch, but the ongoing U.S. trials will inform whether the assistant can scale beyond the test group and become a staple of Meta’s shopping ecosystem.