Meta Enters AI Shopping Race

The tool surfaces product suggestions in a scrollable carousel format, complete with images, brands, and prices. For each recommendation, the AI provides a brief, bullet-pointed explanation of why the specific item was chosen, and allows users to ask for detailed comparisons on specifications, price ranges, and features. While Meta's Llama models are the foundation, some initial queries are reportedly being routed through Google's Gemini 3 during the test phase. The final version is expected to run on Meta's proprietary model, codenamed "Avocado," which is anticipated in the first half of 2026 and will focus on advanced reasoning. This feature is an extension of Meta's broader "Business AI" initiative, which aims to provide 24/7 AI sales agents for retailers. It complements existing generative AI tools in the Advantage+ suite that already create ad variations, generate backgrounds, and expand images to fit different formats like Reels or Feeds. The move targets the rapidly growing generative AI in e-commerce market, valued at over $962 million in 2025 and projected to expand at a CAGR of 15.17% through 2035. The stakes are high, as AI-driven recommendations already account for an estimated 35% of all consumer purchases on Amazon. Building such a system involves more than just the LLM; it requires a robust data pipeline for user interactions (clicks, purchases) and item metadata. The core of these systems often uses a hybrid of collaborative filtering, based on similar user behaviors, and content-based filtering to make predictions. A low-latency API is then needed to serve these recommendations in real-time. For an ML engineering portfolio, a standout project would be to build an end-to-end e-commerce recommendation agent. This could involve fine-tuning an open-source model like Llama 3 on a product dataset, using a Retrieval-Augmented Generation (RAG) pipeline with a vector database for factual grounding, and deploying the system as a containerized service. Top tech companies hiring new-grad ML engineers prioritize this blend of software engineering and production awareness. Resumes stand out when they demonstrate experience with the full ML lifecycle, including data processing, model deployment, MLOps, and cloud infrastructure like Kubernetes for scaling.