Research Report 6: Agentic Shopping is Complicated and Contingent

This is the sixth in a series of short reports that help business, education, and policy leaders understand the technical details of working with AI through rig

A guide to the anatomy of effective commerce agents | Claude by Anthropic

The architecture, latency & cost techniques, and eval practices for agents that make it easier to buy and sell online.

What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commerce

Online marketplaces will be transformed by autonomous AI agents acting on behalf of consumers. Rather than humans browsing and clicking, AI agents can parse webpages or leverage APIs to view, evaluate and choose products. We investigate the behavior of AI agents using ACES, a provider-agnostic framework for auditing agent decision-making. We reveal that agents can exhibit choice homogeneity, often concentrating demand on a few ``modal'' products while ignoring others entirely. Yet, these preferences are unstable: model updates can drastically reshuffle market shares. Furthermore, randomized trials show that while agents have improved over time on simple tasks with a clearly identified best choice, they exhibit strong position biases -- varying across providers and model versions, and persisting even in text-only "headless" interfaces -- undermining any universal notion of a ``top'' rank. Agents also consistently penalize sponsored tags while rewarding platform endorsements, and sensitivities to price, ratings, and reviews vary sharply across models. Finally, we demonstrate that sellers can respond: a seller-side agent making simple, query-conditional description tweaks can drive significant gains in market share. These findings reveal that agentic markets are volatile and fundamentally different from human-centric commerce, highlighting the need for continuous auditing and raising questions for platform design, seller strategy and regulation.

What Is Your AI Agent Buying? Evaluation, Biases, Model Dependence, & Emerging Implications for Agentic E-Commerce

Online marketplaces will be transformed by autonomous AI agents acting on behalf of consumers. Rather than humans browsing and clicking, AI agents can parse webpages or leverage APIs to view, evaluate and choose products. We investigate the behavior of AI agents using ACES, a provider-agnostic framework for auditing agent decision-making. We reveal that agents can exhibit choice homogeneity, often concentrating demand on a few ``modal'' products while ignoring others entirely. Yet, these preferences are unstable: model updates can drastically reshuffle market shares. Furthermore, randomized trials show that while agents have improved over time on simple tasks with a clearly identified best choice, they exhibit strong position biases -- varying across providers and model versions, and persisting even in text-only "headless" interfaces -- undermining any universal notion of a ``top'' rank. Agents also consistently penalize sponsored tags while rewarding platform endorsements, and sensitivities to price, ratings, and reviews vary sharply across models. Finally, we demonstrate that sellers can respond: a seller-side agent making simple, query-conditional description tweaks can drive significant gains in market share. These findings reveal that agentic markets are volatile and fundamentally different from human-centric commerce, highlighting the need for continuous auditing and raising questions for platform design, seller strategy and regulation.

commerce-agents/shopping-agent/core/shopping_agent/gates.py at fd4d59224ab96b43c6dc6888207c67b3bd5a24cf · anthropics/commerce-agents

Reference blueprint for building shopping and merchant agents with Claude. Examples in retail, commerce, telecom, and entertainment included. - anthropics/commerce-agents

Platform Information Provision and Consumer Search: A Field Experiment

Despite substantial efforts to help consumers search in more intuitive ways, text search remains the predominant tool for product discovery online. In this paper, we explore the effects of visual and textual cues for search refinement on consumer search and purchasing behavior. We collaborate with one of the largest e-commerce platforms in China and study its roll out of a new search tool. When a customer searches for a general term (e.g., “headphones”), the tool suggests refined queries (e.g., “bluetooth headphones” or “noise-canceling headphones”) with the help of images and texts. The search tool was rolled out with a long-run experiment, which allows us to measure its short-run and long-run effects. We find that, although there was no immediate effect on orders or total expenditures, the search tool changed customers’ search and purchasing behavior in the long-run. Customers with access to the new tool eventually increased orders and expenditures compared to those in the control group, especially for non top-selling products. The purchase increase comes from more effective searches, rather than an increase in activity on the platform. We also find that the effect is not only driven by the direct value of suggested searches, but also by customers indirectly learning to perform more effective searches on their own.

AI is becoming a shopping channel: report

And 34% of shoppers now use the tech

Agents enter distribution - Sensemaker

Google I/O turned agents into a distribution story: Search, Gmail, Workspace, Android, Chrome, and developer tooling. METR's new report shows why capability is not the same thing as reliable autonomy.

commerce-agents/merchant-agent/core/merchant_agent/gates.py at fd4d59224ab96b43c6dc6888207c67b3bd5a24cf · anthropics/commerce-agents

Reference blueprint for building shopping and merchant agents with Claude. Examples in retail, commerce, telecom, and entertainment included. - anthropics/commerce-agents

Agents First

Every product is getting a second customer — the human who pays, and the agent who decides. A design framework for building products that AI agents can use as primary consumers.

The Shape of Memory Benchmarks

Why the familiar memory benchmarks are outdated, how the agent-native work looks today and why design your own.

LukeW | Agent Management Interface Patterns

As an increasing number of AI applications evolve to agents doing work for people, agent management becomes a critical part of these product's design. How can p...

commerce-agents/docs/safety.md at fd4d59224ab96b43c6dc6888207c67b3bd5a24cf · anthropics/commerce-agents

Reference blueprint for building shopping and merchant agents with Claude. Examples in retail, commerce, telecom, and entertainment included. - anthropics/commerce-agents

Building napkin - a memory system for agents

A decade of information retrieval and three years of agent harness engineering, poured into a local-first knowledge system that avoids vector search entirely.