Alibaba Cloud Healthcare China Translated from Chinese

Yingkang Life (盈康一生)

Enterprise Knowledge Search · IT Operations Automation

Yingkang Life needed to move agents from demonstrations into reliable production for a health ecosystem serving more than 13,000 employees. It needed stable quality, scale, cost control, auditability, and clear permission boundaries so agents could participate safely in medical and organizational workflows.

465 AI-native applications incubated
7,721 high-quality conversation scenarios accumulated in nearly 3 months
Estimated 30.97% reduction in model-call costs
More than 13,000 employees served

Reported impact

Business outcome

The platform became a shared AI entry point for over 13,000 employees. In nearly three months, employees created 465 AI-native applications; the company estimates model-routing improvements can reduce model-call costs by 30.97%.

  • Cost Reduction
  • Productivity Improvement
  • Scalability

Solution

Yingkang Life built Xiaoying Agent OS on Alibaba Cloud's Agentic Cloud foundation, using AgentCore for agent execution and governance and AgentLoop for observation and optimization. The AI Gateway handles authentication and routing; Bailian provides multimodal foundation models; Supabase, APIs, and MCP connect business systems, data, and tools; and IAM establishes user-agent identity, least privilege, and revocable authorization. AgentCore supplies isolated sandboxes, multi-tenant workspaces, orchestration, and release gates, while AgentLoop traces tasks and feeds failures and successful runs into evaluation and review. Agents use data and capability maps to collaborate in approved applications, with human review, rollback, and audit trails. The platform is used as a shared employee AI entry point for office work, application creation, and coding.

Data flow

Business information and tools connect through Supabase, APIs, and MCP; requests pass through the AI Gateway to models and agents running in AgentCore. AgentLoop records traces and evaluation results, while IAM constrains access and human approval gates changes before release.

Solution architecture

7 components · 4 layers
  1. Compute
    • Bailian large model service Supplies the large and multimodal models used by the enterprise agents.
  2. Orchestration
    • AgentCore Runs and orchestrates agents in isolated, multi-tenant environments with versioning and release controls.
    • Alibaba Cloud AI Gateway Provides unified authentication and model routing for agent requests.
  3. Governance
    • AgentLoop Traces agent execution and supports evaluation, diagnosis, and iterative improvement.
    • IAM Controls user and agent identities, permissions, and revocation.
  4. Uncategorized
    • Supabase Connects business data and systems to the agent applications.
    • Qoder Supports employee code creation and collaboration through the shared AI entry point.

Architecture clues

  • Agent execution uses millisecond-level isolated sandboxes and remote workspaces
  • Release gates support human confirmation, rollback, and audit
  • User and agent identities use least-privilege authorization and dynamic credentials

Evidence from the source

AgentCore and AgentLoop form a feedback loop of use observation, evaluation and review, and iteration on new requirements (original: AgentCore × AgentLoop 形成“使用观测 → 评估复盘 → 新需求迭代”的反馈回路).
The employee is transformed from an AI user into an application co-creator (original: 员工从 AI 使用者转变为应用共创者).
Transparency Evidence & methodology 90/100 reference value · 83% confidence · 2 sources

Review how this reference was scored and trace each normalized claim back to its documented source.

01

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value90/100
  • Problem clarity20/20
    Measures whether the customer's operational problem is specific enough to compare.
  • Implementation depth25/25
    Rewards named components, data flow, integrations, and architecture detail.
  • Outcome specificity20/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact15/25
    1 metric describe measurable impact.
  • Deployment scale10/10
    Rewards evidence of a live or scaled deployment and meaningful integration scope.
Evidence confidence83%
  • Source authority20/25
    The main evidence is an official vendor-published customer story.
  • Claim traceability23/25
    11 of 12 key fields link to a named source.
  • Direct evidence10/20
    2 direct source quotes retained.
  • Date provenance10/10
    The publication date comes from the source page or publisher.
  • Source transparency10/10
    2 source references document publisher, URL, authority, retrieval, and supported fields.
  • Implementation specificity10/10
    Rewards source-backed components, data flow, quantified impact, and a specific outcome.
02

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

  1. Alibaba Cloud

    Yingkang Life and Alibaba Cloud build an enterprise Agent OS

    First-party
    https://www.aliyun.com/customer-stories/health-care-2026-haier
    Source type
    Vendor-published customer story
    Supports
    • Customer
    • Industry
    • Business problem
    • Solution
    • Products
    • Solution components
    • Data flow
    • Integrations
    • Business outcome
    • Metrics
    • Architecture
  2. Yingkang Life

    About Yingkang Life

    First-party
    https://www.yingkanglife.net/about

    Yingkang Life identifies itself as a healthcare ecosystem brand established by Haier Group and describes its operating footprint across China’s Bohai Rim, Yangtze River Delta, Greater Bay Area, and Chengdu–Chongqing regions.

    Source type
    Official company website
    Supports
    • Region

Date evidence

  • Published on source page Nov 9, 2026 · Alibaba Cloud · First-party
    Open date source
Not separately sourced Company size. Treat these normalized attributes as lower-confidence until a dedicated source is added.