Amazon launches AI agent to help sellers complete tasks and manage their businesses (techcrunch.com)

Amazon updated Seller Assistant into an always-on, agentic AI that proactively helps third-party sellers run operations, monitor account health, and — when explicitly authorized — take actions on their behalf. The upgraded assistant continuously scans inventory and account signals, flags slow-moving SKUs before long-term storage fees, analyzes demand patterns to recommend pricing or removals, prepares shipment recommendations, and checks listings for cross-border compliance with product-safety rules. Amazon also extended agentic capabilities into ads, letting sellers generate campaigns through conversational prompts, and positions the assistant to move beyond routine tasks into strategic decision support. For the AI/ML community this marks a concrete push toward agent-driven commerce: models tied into seller data and operational pipelines can automate end-to-end workflows, from forecasting and repricing to compliance enforcement. Technically, that requires continuous data ingest, demand-forecast models, policy-compliance classifiers across jurisdictions, and secure action authorization/rollback controls. The move amplifies efficiency gains for sellers but also raises integration and safety questions — auditability, human-in-the-loop safeguards, permissioning, and payment/transaction protocols (Google recently published a protocol for agentic payments). Seller Assistant demonstrates how ML agents are shifting from assistive tools to orchestrators of real-world business processes, creating new product and governance challenges for ML engineers and platform architects.