百秋尚美集团发布NexxAI无人值守数字零售运营智能体生态 实现Copilot到Autopilot跃迁

Core takeaway: Baqiu Shangmei Group has launched NexxAI, an unattended digital retail operations agent ecosystem. Chairman and CEO Liu Zhicheng argues that e-commerce operations will move from Copilot to Autopilot.

1. Clear roadmap: L1 assisted generation, L2 full execution, L3 autonomous decision-making, and L4 unattended operations, providing a benchmark for assessing stages of AI application.

2. Deployment scenarios: starting with livestreaming, offering second-level real-time response, minute-level livestream decisions, and hour-level learning and review, then expanding to product, content, media buying, operations, and customer service.

3. Operational support: the proprietary TAMC multimodal temporal fact storage engine can replay any livestream; RSI recursive self-iteration enables system self-evolution; System One/System Two fast and slow thinking balance immediate response and long-cycle reasoning.

4. Performance reference: tests show that effective decisions, decision reference dimensions, and concurrent task processing reach several times human levels, while driving GMV and order growth.

Baqiu has packaged years of digital retail operating experience into the NexxAI agent ecosystem, directly serving brands’ daily operations—a change brands should watch most closely.

1. Operational collaboration: agents for different operational stages collaborate around brand goals, understand the operating context, identify problems and opportunities, form action plans, and execute and review improvements under clear permissions and human oversight.

2. Service value: brands can obtain timely, stable operating services aligned with their own goals; agents participate in problem discovery, coordinated response, and performance review, continuously improving service quality.

3. Growth reference: tests show that NexxAI delivers value in decision quality, demand response, and revenue growth. Effective decisions per livestream, reference dimensions per decision, and batch task processing reach several times human levels, while driving positive growth in livestream GMV and order volume.

4. Partnership signal: at the launch event, brand clients showed strong interest in the features and expected broader collaboration; as efficiency improves, NexxAI may also serve more small and midsize brands with integrated solutions.

For sellers, NexxAI signals a shift in e-commerce operations from competing on traffic and manpower to competing on operating quality, response efficiency, and continuous service capability.

1. New model: Baqiu proposes an L1-to-L4 path for unattended stores, with NexxAI targeting Autopilot so AI can autonomously complete the closed loop of perception, decision-making, execution, and review—and be accountable for operating results.

2. What to learn: the point is not merely personal productivity gains but reconstructing processes around business scenarios, replacing people at process nodes with AI, and building an AI-native organization; it also stresses that unattended does not mean unaccountable, requiring clear permissions and human oversight.

3. Opportunity signals: start with livestreaming, providing second-level response, minute-level decisions, and hour-level review, then expand to product, content, media buying, operations, and customer service; in tests, effective decisions and concurrent task processing reach several times human levels, with positive growth in GMV and order volume.

4. Collaboration: brand clients can work with Baqiu in more NexxAI areas, and smaller brands are expected to gain access to integrated solutions in the future.

Although the article does not directly address factory production, it offers several implications for digitalization and e-commerce operations.

1. Organizational digitalization: Baqiu emphasizes an AI-native organization—not using point tools to improve individual productivity, but mapping the overall workflow and focusing on business scenarios to replace people at process nodes with agents. This has reference value for factory process reengineering.

2. System foundation: Baqiu’s proprietary Futail retail system provides a system foundation for agent execution; NexxAI is based on real business scenarios and long-term operating experience, showing that digitalization requires both business systems and accumulated data.

3. Technology productization: the livestreaming version features the proprietary TAMC multimodal temporal fact storage engine, integrating cross-modal information along a timeline to support replay and review; the RSI recursive self-iteration mechanism feeds decision trajectories and feedback back into optimization; System One/System Two fast and slow thinking balance immediate judgment and long-cycle planning.

4. Business opportunities: NexxAI is expanding from livestreaming to product, content, media buying, operations, and customer service, and may serve more small and midsize brands in the future, indicating ecosystem collaboration and solution opportunities in digital retail operations tools.

For service providers, NexxAI demonstrates a trend in digital retail services: moving from individual experience-driven work to collaboration between professional teams and agents.

1. Client pain points: the decisive factor in e-commerce operations is shifting from human experience to organizational capability. Manpower alone struggles to see all data simultaneously, monitor all hours, and handle concurrent demand, while stores easily become dependent on individual experience.

2. Solution: based on Baqiu’s long-term operating experience, real business scenarios, and system capabilities, NexxAI enables agents across different stages to collaborate around brand goals—understanding the context, identifying problems and opportunities, forming plans, and executing and reviewing improvements under permissions and oversight.

3. New technologies: the livestreaming version features the proprietary TAMC multimodal temporal fact storage engine, using a timeline anchor to integrate cross-modal information and support replay of any livestream; the RSI recursive self-iteration mechanism feeds decision trajectories and feedback back for continuous optimization; System One/System Two fast and slow thinking balance low-latency response with long-cycle planning and reasoning.

4. Trend and results: AI is entering a phase of deep integration with real business. Tests show that effective decisions, reference dimensions, and batch task processing reach several times human levels, while driving GMV and order growth.

Marketplace operators can draw from NexxAI lessons about operations management and ecosystem collaboration as AI agents enter retail operations.

1. Operations management: Baqiu proposes an L1-to-L4 path for unattended stores, aiming for AI to complete the closed loop of perception, decision-making, execution, and review and to be accountable for operating results; however, unattended does not mean unaccountable—clear permissions and human oversight are required, offering a reference for platforms designing risk-control boundaries.

2. Merchant services: NexxAI’s agents for different operational stages collaborate around brand goals, understand the operating context, identify problems and opportunities, form action plans, and review improvements—useful as a capability reference for platforms serving brand merchants.

3. Technology foundation: the proprietary Futail retail system provides the system foundation for agent execution; the TAMC multimodal temporal fact storage engine supports full-chain fact capture and replay for livestreaming; the RSI recursive self-iteration mechanism drives system self-evolution; System One/System Two fast and slow thinking balance real-time response and long-term planning.

4. Ecosystem signals: the launch event attracted brands, investors, and media representatives, with Sequoia China partner Liu Xing participating in the ribbon-cutting; brand clients look forward to broader collaboration, and NexxAI may serve more small and midsize brands in the future.

Researchers can view NexxAI as an industry sample of AI application in digital retail.

1. New development: Baqiu Shangmei proposes an L1-to-L4 path for unattended stores—from assisted generation and full execution to autonomous decision-making and unattended operations—and distinguishes Copilot tool assistance from Autopilot autonomous closed loops, providing a benchmark for assessing stages of e-commerce AI application.

2. New questions: the article argues that AI application should not focus only on individual productivity but should build AI-native organizations around the overall workflow; it also stresses that unattended does not mean unaccountable, requiring clear permissions and human oversight, which involves organizational restructuring and responsibility boundaries.

3. Technical route: the livestreaming version features the proprietary TAMC multimodal temporal fact storage engine, using a timeline anchor to integrate cross-modal information; the RSI recursive self-iteration mechanism feeds decision trajectories and feedback back for optimization; System One/System Two fast and slow thinking coordinate immediate response and long-cycle reasoning.

4. Business model: NexxAI productizes Baqiu’s long-term operating experience and enters brands’ daily operations through an agent ecosystem, moving from individual experience-driven work to collaboration between professional teams and agents; tests show value in decision-making, response, and revenue growth, and it may serve more small and midsize brands in the future, representing a new exploration in digital retail services.