Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note4 min read

ShopX: A Foundation Model for Intent-to-Item Fulfillment in Agentic Shopping

Source

arXiv

arxiv.org/abs/2606.31693v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Jiacheng Chen, Tao Zhang, Manxi Lin, Dunxian Huang, Teng Shi, Honghao Fu, Mengyan Li, Xinming Zhang, Chenchi Zhang, Xuan Lu, Xiaoxiong Du, Haibin Chen, Shaolin Ye, Hao Chang, Xiaoqi Li, Shuwen Xiao, Yujin Yuan, Jingxuan Feng, Shaopan Xiong, Huimin Yi, Ju Huang, Qiu Shen, Ying Chen, Junjun Zheng, Xiangheng Kong, Yuning Jiang
  • Venue: arXiv: cs.IR
  • Published: 2026-06-30
  • Collected: 2026-07-01
  • URL: http://arxiv.org/abs/2606.31693v1
  • PDF: https://arxiv.org/pdf/2606.31693v1
  • DOI: UNKNOWN

Plain-Language Brief

The paper proposes a foundation model named ShopX to enhance AI shopping applications by bridging the gap between language understanding and item-space fulfillment. The model allows LLMs (large language models) to interact directly with items in the catalog using semantic IDs (SIDs), and it can plan and execute various operations such as retrieval, ranking, or product bundling. The authors claim that ShopX improves overall framework behavior in agentic shopping workflows, particularly on complex or ambiguous requests, compared to tool-mediated systems.

What the Document Says

  • The AI-native shopping applications are shifting towards intent-driven experiences managed by LLM agents.
  • Current designs wrap an LLM around existing search and recommendation pipelines, which can result in complex intents being forced through low-bandwidth interfaces.
  • Generative recommendation provides LLMs with a direct item-space interface via SIDs but mainly generates candidates for retrieval rather than translating flexible intents into item-space outcomes.
  • The authors propose ShopX to address this bottleneck by integrating intent understanding, execution planning, and flexible SID-native item-space operations.
  • ShopX is deployed in agentic shopping workflows through a model-native item-fulfillment framework.
  • The framework defines a model-facing action protocol and exposes support surfaces for context access, catalog grounding, and state management.
  • Within the framework, ShopX plans and composes SID-based item-space operations like SID beam-search retrieval, listwise ranking, or product bundling.
  • The design aims to reduce lossy hand-offs between agent orchestration and item-space execution.
  • The authors design semantically recoverable, LLM-operable SIDs and a training recipe for the model.
  • The general LLM is equipped for flexible multi-turn item-space fulfillment while retaining the knowledge and instruction-following abilities needed by a shopping agent.
  • The evaluation of ShopX framework is against tool-mediated agentic systems on single- and multi-turn fulfillment tasks derived from Taobao production logs.
  • The authors claim that model-native fulfillment improves overall framework behavior, particularly on complex or ambiguous requests.

Key Concepts Mentioned

  • AI shopping applications
  • LLM agents
  • intent-driven experiences
  • semantic IDs (SIDs)
  • generative recommendation
  • ShopX foundation model
  • model-native item-fulfillment framework
  • catalog grounding
  • state management
  • single-turn fulfillment tasks
  • multi-turn fulfillment tasks

Claims Made by the Document

  • Existing AI shopping applications have a gap between language understanding and item-space fulfillment.
  • ShopX can address the bottleneck by unifying intent understanding, execution planning, and flexible SID-native item-space operations into a single foundation model.
  • ShopX improves overall framework behavior in agentic shopping workflows compared to tool-mediated systems.
  • Model-native fulfillment particularly improves performance on complex or ambiguous requests.

Evidence Used

  • The wave of AI-native applications is moving shopping beyond page- and feed-based browsing towards intent-driven experiences orchestrated by LLM agents
  • Generative recommendation gives LLMs a direct item-space interface through semantic IDs (SIDs), but existing models mainly generate candidates for retrieval rather than translate flexible intents into item-space outcomes.
  • To build ShopX, we design semantically recoverable, LLM-operable SIDs and a training recipe that equips a general LLM for flexible multi-turn item-space fulfillment while retaining the knowledge and instruction-following abilities needed by a shopping agent.
  • We evaluate the ShopX framework against tool-mediated agentic systems on single- and multi-turn fulfillment tasks derived from anonymized Taobao production logs.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

The wave of AI-native applications is moving shopping beyond page- and feed-based browsing toward intent-driven experiences orchestrated by LLM agents. A common design wraps an LLM around existing search and recommendation pipelines, forcing complex intents through low-bandwidth retrieval or ranking interfaces and leaving a gap between language understanding and item-space fulfillment. Generative recommendation gives LLMs a direct item-space interface through semantic IDs (SIDs), but existing models mainly generate candidates for retrieval rather than translate flexible intents into item-space outcomes. We propose ShopX to address this bottleneck by unifying intent understanding, execution planning, and flexible SID-native item-space operations into a single foundation model. We deploy ShopX in agentic shopping workflows through a model-native item-fulfillment framework with a serving harness that defines a model-facing action protocol and exposes support surfaces for context access, catalog grounding, and state management. Within this framework, ShopX plans and composes SID-based item-space operations such as SID beam-search retrieval, listwise ranking, or product bundling. This model-centric design reduces lossy hand-offs between agent orchestration and item-space execution. To build ShopX, we design semantically recoverable, LLM-operable SIDs and a training recipe that equips a general LLM for flexible multi-turn item-space fulfillment while retaining the knowledge and instruction-following abilities needed by a shopping agent. We evaluate the ShopX framework against tool-mediated agentic systems on single- and multi-turn fulfillment tasks derived from anonymized Taobao production logs, showing that model-native fulfillment improves overall framework behavior, especially on complex or ambiguous requests.