By Interestana AI Editorial — AI-drafted, human-overseen. How we report
Anthropic Releases Apache-2.0 Commerce Agent Blueprint
Anthropic released anthropics/commerce-agents this week, an open-source blueprint designed to streamline the development of AI agents for e-commerce and customer service. This initiative addresses the common challenge faced by development teams who often rebuild foundational components for shopping assistants, such as agent loops, tool layers for product catalogs, approval mechanisms, and evaluation suites. The released blueprint provides this scaffolding as ready-to-use code, aiming to accelerate the deployment of sophisticated shopping and merchant agents across various industries.
The blueprint includes two primary agents: a shopping agent and a merchant agent. The shopping agent is designed to operate within a merchant's application, capable of searching product catalogs, managing multi-item requests, comparing options, building shopping carts, and handling customer inquiries regarding orders and returns within a single conversational thread. Its functionalities are categorized into five core skills: search-discovery, purchase-research, planning-goals, customer-care, and memory-personalization. For deployment, it requires a StorefrontBackend that interfaces with the merchant's catalog, cart, order, and policy systems.
The merchant agent is intended to support store staff by providing insights into sales performance, issuing inventory alerts, recommending pricing and promotion strategies, and assisting with the drafting of marketing campaigns. Its six skills include performance-insights, catalog-listings, inventory-operations, pricing-promotions, and marketing-campaigns, operating over a MerchantBackend. Both agents can be run in three distinct modes: via the Messages API, the Claude Agent SDK, and Claude Managed Agents (in beta). This unified approach allows for a single definition of prompts, skills, tool contracts, and approval gates to be used across all execution environments.
Anthropic's architectural approach emphasizes the development of distinct skills rather than relying on sub-agents, which they argue is a more transferable and efficient design. The repository is licensed under the Apache 2.0 license, making it freely available for commercial and non-commercial use. It is compatible with Python 3.11+ and Node 22, requiring an ANTHROPIC_API_KEY for operation. The runtimes are designed to be flexible, accepting any Anthropic client, which enables deployment across major cloud platforms including the Claude API, Amazon Bedrock, Microsoft Foundry, and Google Cloud Vertex AI. Additionally, a Claude Code plugin, commerce-builder, is provided to scaffold new agents or review existing ones, further simplifying the development and management process for these commerce-focused AI applications. The blueprint supports four specific verticals: retail, travel, telecom, and entertainment, demonstrating its adaptability to diverse business needs within the e-commerce landscape.
Original source — read the full reporting at the publisher:
Read on MarkTechPostGet the weekly AI digest
AI news + new model releases, weekly. Drafted by our agents, reviewed by humans.