# TalentVerse: Dual-Track Talent Intelligence Platform
# TalentVerse: 双轨人才智能平台
*Internal Architecture Proposal · TPC/OCTAVE · April 2026*
*内部架构提案 · TPC/OCTAVE · 2026年4月*

**Classification**: Confidential — Internal Only
**分类**：机密 — 仅限内部

**Status**: Awaiting Pre-Build Gate Approval
**状态**：等待预构建门控审批
**Prepared for**: TPC/OCTAVE Leadership (TPC Board / OCTAVE Executive Team)
**编制对象**：TPC/OCTAVE 领导层（TPC 董事会 / OCTAVE 执行团队）
**Delivery target**: October 15, 2026
**交付目标**：2026年10月15日

---

## Table of Contents
## 目录

1. [Executive Summary](#1-executive-summary)
1. [执行摘要](#1-执行摘要)
2. [Solution: TalentVerse Dual-Track Talent Intelligence](#3-solution-talenverse-dual-track-talent-intelligence)
2. [解决方案：TalentVerse 双轨人才智能](#3-解决方案talenverse-双轨人才智能)
3. [Scope and Boundaries](#4-scope-and-boundaries)
3. [范围与边界](#4-范围与边界)
4. [Data Architecture](#5-data-architecture)
4. [数据架构](#5-数据架构)
5. [Development Framework Layer (Track B)](#6-development-framework-layer-track-b)
5. [发展框架层（B 轨）](#6-发展框架层b-轨)
6. [Stage-Based 2026 Roadmap](#7-stage-based-2026-roadmap)
6. [基于阶段的 2026 路线图](#7-基于阶段的-2026-路线图)
7. [Sprint Plan Summary](#8-sprint-plan-summary)
7. [冲刺计划摘要](#8-冲刺计划摘要)
8. [What Success Looks Like: October 15, 2026](#14-what-success-looks-like-october-15-2026)
8. [成功标准：2026年10月15日](#14-成功标准2026年10月15日)
9. [Phase 2 and Beyond](#15-phase-2-and-beyond)
9. [第二阶段及以后](#15-第二阶段及以后)

---

## 1. Executive Summary
## 1. 执行摘要

### Situation
### 现状

OCTAVE Institute manages 240 employees across a talent lifecycle spanning onboarding, performance, development, governance participation, and transitions. TPC's mandate, reflected in the RMRR framework and M5 leadership principles, requires systematic development and retention of high-potential talent. That mandate requires visibility into who is performing, who is growing, and at what velocity.
OCTAVE 研究所管理着跨越人才生命周期的 240 名员工，包括入职、绩效、发展、治理参与和过渡。TPC 的任务，体现在 RMRR 框架和 M5 领导力原则中，要求系统地发展和保留高潜力人才。该任务需要对谁在表现、谁在成长以及以什么速度成长的可见性。

Today, neither type of visibility exists in any single place. Performance data is fragmented across six disconnected systems. Development signals — coaching conversations, programme participation, reflections, 1-on-1 transcripts — are captured in unstructured formats with no longitudinal synthesis. Every talent decision is made on partial evidence, assembled manually, assessed subjectively.
今天，这两种可见性都不存在于任何一个地方。绩效数据分散在六个不连接的系统中。发展信号——教练对话、项目参与、反思、一对一转录——以非结构化格式捕获，没有纵向综合。每个人才决策都是在部分证据、手动组装、主观评估的基础上做出的。

### Problem
### 问题

OCTAVE faces two connected talent intelligence gaps. The first is a performance evidence gap: no single system holds multi-source longitudinal performance evidence, so calibration defaults to recency bias, halo effects, and manager advocacy. Employees with strong trajectories but low political visibility are systematically disadvantaged.
OCTAVE 面临两个相互关联的人才智能差距。第一个是绩效证据差距：没有单一系统保存多源纵向绩效证据，因此校准默认为近因偏见、光环效应和经理倡导。具有强劲轨迹但政治可见度低的员工处于系统性劣势。

The second is a development visibility gap: OCTAVE currently has no mechanism to answer whether a person is growing, and in what direction. Development conversations remain episodic. Growth investments cannot be connected to behavioral change. Promotional readiness assessments rely on intuition.
第二个是发展可见性差距：OCTAVE 目前没有办法回答一个人是否在成长，以及向什么方向成长。发展对话仍然是偶发的。成长投资无法与行为改变联系起来。晋升准备评估依赖于直觉。

Both problems must be solved together for TalentVerse to fulfill its stated purpose.
这两个问题必须一起解决，TalentVerse 才能实现其既定目标。

### Solution
### 解决方案

TalentVerse is a dual-track talent intelligence platform:
TalentVerse 是一个双轨人才智能平台：

**Track A — Performance Intelligence.** Multi-source, evidence-backed assessment across five performance dimensions (D1-D5) and three manager-effectiveness dimensions (M1-M3). Medallion architecture (Bronze, Silver, Gold). SHAP/LIME explainability for every insight. Four-tier PDPA consent architecture enforced from ingestion to inference.
**A 轨 — 绩效智能。** 跨五个绩效维度（D1-D5）和三个管理者效能维度（M1-M3）的多源、证据支持的评估。奖章架构（青铜、白银、黄金）。每个洞察都有 SHAP/LIME 可解释性。从摄取到推理强制执行四层 PDPA 同意架构。

**Track B — Development Intelligence.** Nine development framework profiles inferred from unstructured sources (1-on-1 transcripts, coaching reports, programme participation, reflection journals) using an LLM inference pipeline. Four growth dimensions: G1 Self-Cultivation (修身, Xiushen), G2 Roles and Relationships (齐家, Qijia), G3 Stewardship (治国, Zhiguo), G4 Consciousness (平天下, 天人合一, Ping tianxia, Tian ren he yi). Two derived composites: LVTS (Longitudinal Value Trajectory) and PRCS (Promotional Readiness Confidence).
**B 轨 — 发展智能。** 使用 LLM 推理管道从非结构化来源（一对一转录、教练报告、项目参与、反思日志）推断的九个发展框架概况。四个成长维度：G1 自我修养（修身）、G2 角色与关系（齐家）、G3 管理（治国）、G4 意识（平天下、天人合一）。两个派生综合指标：LVTS（纵向价值轨迹）和 PRCS（晋升准备信心）。

Both tracks operate on the same Medallion infrastructure and the same Employee Digital Twin model. Track B data is protected by a dedicated data firewall that enforces role-based access with stricter controls than Track A.
两轨都在相同的奖章基础设施和相同的员工数字孪生模型上运行。B 轨数据受到专用数据防火墙的保护，执行比 A 轨更严格的基于角色的访问控制。

### October 15, 2026 Target
### 2026年10月15日目标

The October 15, 2026 go-live target is confirmed. SMP Pillars are not included in Track B, which reduces pipeline complexity in G3. Mintzberg Mindset applies to all employees, broadening inference coverage within the same wave structure. Neither decision adds net delivery risk.
2026年10月15日上线目标已确认。SMP 支柱不包括在 B 轨中，这减少了 G3 中的管道复杂性。明茨伯格思维模式适用于所有员工，在相同的波浪结构中扩大推理覆盖范围。这两项决定都不会增加净交付风险。

### Investment
### 投资

The resource plan is fully scoped and confirmed. SMP removal reduces Track B inference workload. The Mintzberg Mindset expansion reuses the same prompt engineering work already planned. No net headcount change is required.
资源计划已完全确定范围并确认。SMP 移除减少了 B 轨推理工作量。明茨伯格思维模式扩展重用已计划的相同提示工程工作。不需要净人员编制变更。


## 3. Solution: TalentVerse Dual-Track Talent Intelligence
## 3. 解决方案：TalentVerse 双轨人才智能

### 3.1 What TalentVerse Does
### 3.1 TalentVerse 做什么

TalentVerse ingests data from multiple source systems, transforms it into structured Evidence Objects, and produces two parallel intelligence layers for every employee:
TalentVerse 从多个源系统摄取数据，将其转换为结构化的证据对象，并为每位员工产生两个平行的智能层：

- **Track A**: A multi-source dimensional insight (D1-D5 for all employees, M1-M3 for people managers) backed by an Evidence Trail, a confidence level, and a SHAP/LIME attribution showing which sources drove the insight.
- **A 轨**：多源维度洞察（所有员工的 D1-D5，人员管理者的 M1-M3），由证据轨迹、置信度水平和 SHAP/LIME 归因支持，显示哪些来源推动了洞察。
- **Track B**: Nine development framework profiles inferred from unstructured data using an LLM pipeline, plus two derived composites (LVTS and PRCS) available at Wave 6.
- **B 轨**：使用 LLM 管道从非结构化数据推断的九个发展框架概况，加上两个派生综合指标（LVTS 和 PRCS），在第 6 波可用。

Both tracks populate the Employee Digital Twin: a continuously updated longitudinal record of an employee's performance and development trajectory across full tenure.
两轨都填充员工数字孪生：员工整个任期内绩效和发展轨迹的持续更新的纵向记录。

### 3.2 Medallion Architecture
### 3.2 奖章架构

TalentVerse runs on Microsoft Azure Southeast Asia (Singapore) for PDPA data residency compliance.
TalentVerse 运行在 Microsoft Azure 东南亚（新加坡）上，以符合 PDPA 数据驻留要求。

**Bronze layer.** Immutable raw storage. Every record from every source system, exactly as received, tagged with a consent tier. A KIV Classification Filter at Bronze routes unvalidated instruments (Aura Photography, TimeWaver, Hawkins Consciousness Scale, 奇门遁甲) to /bronze/kiv/ before any Silver processing.
**青铜层。** 不可变的原始存储。来自每个源系统的每条记录，完全按接收时的样子，标记同意层级。青铜层的 KIV 分类过滤器将未经验证的仪器（光环摄影、TimeWaver、霍金斯意识量表、奇门遁甲）路由到 /bronze/kiv/，然后进行任何白银处理。

**Silver layer.** Transformation. Every Bronze record becomes a typed, normalised Evidence Object with a canonical employee UUID, a source reliability weight, a consent tier, and an evidence decay half-life.
**白银层。** 转换。每条青铜记录成为类型化的、规范化的证据对象，具有规范的员工 UUID、源可靠性权重、同意层级和证据衰减半衰期。

**Gold layer.** Intelligence. Employee Digital Twins, dimensional insights, development framework profiles, trajectory indicators, and SHAP/LIME explanations. Accessed through a role-gated API.
**黄金层。** 智能。员工数字孪生、维度洞察、发展框架概况、轨迹指标和 SHAP/LIME 解释。通过角色门控 API 访问。

A Track B data firewall operates between the Silver and Gold layers. It enforces the access matrix defined in Section 6.
B 轨数据防火墙在白银层和黄金层之间运行。它执行第 6 节中定义的访问矩阵。

### 3.3 Dual-Track Model
### 3.3 双轨模型

**Track A — Performance Intelligence (D1-D5, M1-M3)**
**A 轨 — 绩效智能（D1-D5，M1-M3）**

| Dimension | What It Measures |
| 维度 | 衡量内容 |
|---|---|
| D1 Delivering Results | Objective achievement, KPI completion, goal execution quality |
| D1 交付成果 | 目标达成、KPI 完成、目标执行质量 |
| D2 Learning and Self-Cultivation | Formal learning, qualifications growth, development behavior |
| D2 学习与自我修养 | 正式学习、资格增长、发展行为 |
| D3 Collaboration and Partnership | Cross-functional engagement, stakeholder relationships, peer effectiveness |
| D3 协作与伙伴关系 | 跨职能参与、利益相关者关系、同伴效能 |
| D4 Stewardship and Accountability | Role responsibility, governance participation, execution discipline |
| D4 管理与问责 | 角色责任、治理参与、执行纪律 |
| D5 Innovation and Ownership | Initiative, entrepreneurial contribution, proactive change |
| D5 创新与所有权 | 主动性、创业贡献、主动变革 |
| M1 Manage with Integrity | Ethical leadership, values-aligned management behavior |
| M1 真诚管理 | 道德领导力、与价值观一致的管理行为 |
| M2 Manage through Coaching | Development investment in direct reports, coaching quality |
| M2 教练式管理 | 对直接下属的发展投资、教练质量 |
| M3 Manage with Connection | Relational investment, care, connection behaviors |
| M3 关系型管理 | 关系投资、关怀、连接行为 |

Each dimensional insight is computed with a confidence weight derived from source count, source quality, and recency. An insight from three or more sources is HIGH confidence. Two sources is MEDIUM. One source is LOW. Only HIGH confidence insights are routinely cited in calibration.
每个维度洞察都使用从源数量、源质量和新近度派生的置信度权重计算。来自三个或更多来源的洞察是高置信度。两个来源是中等。一个来源是低。只有高置信度洞察在校准中常规引用。

**Track B — Development Intelligence (9 frameworks + LVTS + PRCS)**
**B 轨 — 发展智能（9 个框架 + LVTS + PRCS）**

Four growth dimensions of development framework inference, activated progressively by wave. Full detail in Section 6.
按波次逐步激活的四个成长维度发展框架推理。第 6 节有完整详情。

### 3.4 Employee Digital Twin
### 3.4 员工数字孪生

The Employee Digital Twin is a continuously updated longitudinal record. At MVP, it holds both Track A performance evidence and Track B development profiles. The underlying schema covers all six lifecycle stages: Onboarding, Performance, Career Development, Leadership Development, Governance Participation, and Separation. Track B activates the Career Development stage, raising LVTAR because Digital Twins now draw from two stages rather than one.
员工数字孪生是一个持续更新的纵向记录。在 MVP 时，它持有 A 轨绩效证据和 B 轨发展概况。底层模式涵盖所有六个生命周期阶段：入职、绩效、职业发展、领导力发展、治理参与和离职。B 轨激活职业发展阶段，提高 LVTAR，因为数字孪生现在从两个阶段而不是一个阶段提取数据。

Employees have a self-view of their own Digital Twin. They can see both tracks, manage their consent for each data type, and flag evidence they dispute. The Digital Twin is not a surveillance file. It does not hold health data, biometric data, compensation data, or data from unvalidated instruments.
员工拥有自己的数字孪生的自我视图。他们可以看到两轨，管理每种数据类型的同意，并对他们有异议的证据提出标记。数字孪生不是监视文件。它不保存健康数据、生物识别数据、薪酬数据或来自未经验证仪器的数据。

### 3.5 PDPA Compliance by Design
### 3.5 设计即合规 PDPA

Four independent enforcement layers:
四个独立的执行层：

1. **Consent at ingestion.** Every inbound record is tagged at Bronze. Tier 1 (BIPO, OKR) proceeds under employment agreement. Tier 2 (Viva Insights, Teams transcripts) requires explicit consent before flowing beyond Bronze. Tier 3 (SHL psychometrics, MT Journal, Relational Matrix, Consciousness Calibration) requires written explicit consent.
1. **摄取时同意。** 每条入站记录在青铜层标记。一级（BIPO、OKR）在雇佣协议下继续。二级（Viva Insights、Teams 转录）需要明确同意才能在青铜层之外流动。三级（SHL 心理测评、MT 日志、关系矩阵、意识校准）需要书面明确同意。

2. **API-layer enforcement.** Every API endpoint checks the ConsentRecord entity before responding. UI-layer filtering alone is not sufficient.
2. **API 层强制执行。** 每个 API 端点在响应前检查 ConsentRecord 实体。单独的 UI 层过滤是不够的。

3. **Synchronous withdrawal.** Consent withdrawal nulls all insight fields atomically in one database transaction. No window exists during which a manager could retrieve insights from withdrawn evidence.
3. **同步撤回。** 同意撤回在一个数据库事务中原子地清空所有洞察字段。不存在经理可以从撤回证据中检索洞察的窗口。

4. **L1/L2 Firewall.** TPC corporate staff data and OCTAVE employee data are partitioned at the data layer via PostgreSQL Row-Level Security with FORCE ROW LEVEL SECURITY. Cross-tenant data is physically inaccessible.
4. **L1/L2 防火墙。** TPC 公司员工数据和 OCTAVE 员工数据通过 PostgreSQL 行级安全与 FORCE ROW LEVEL SECURITY 在数据层分区。跨租户数据在物理上不可访问。

---

## 4. Scope and Boundaries
## 4. 范围与边界

### 4.1 MVP Scope (October 15, 2026)
### 4.1 MVP 范围（2026年10月15日）

The following capabilities are in scope for the October 15, 2026 MVP:
以下功能在 2026年10月15日 MVP 的范围内：

**Track A (Performance Intelligence)**
**A 轨（绩效智能）**
- All five performance dimensions: D1-D5 for all employees
- 所有五个绩效维度：所有员工的 D1-D5
- All three manager-effectiveness dimensions: M1-M3 for people managers
- 所有三个管理者效能维度：人员管理者的 M1-M3
- Medallion architecture (Bronze, Silver, Gold)
- 奖章架构（青铜、白银、黄金）
- Azure Southeast Asia data residency
- Azure 东南亚数据驻留
- SHAP/LIME explainability for every insight
- 每个洞察都有 SHAP/LIME 可解释性
- KIV classification at Bronze for unvalidated instruments
- 青铜层对未经验证仪器的 KIV 分类
- Four-tier PDPA consent architecture
- 四层 PDPA 同意架构
- Employee Digital Twin schema
- 员工数字孪生模式
- EIDR and LVTAR as North Star metrics
- EIDR 和 LVTAR 作为北极星指标
- L1/L2 firewall enforcement
- L1/L2 防火墙执行
- Synchronous consent withdrawal nullification
- 同步同意撤回清空
- W13 Leadership Demo milestone (Week 8, May 29, 2026)
- W13 领导力演示里程碑（第 8 周，2026年5月29日）

**Track B (Development Intelligence)**
**B 轨（发展智能）**
- Eight development frameworks: 8 Living Habits, TPC Guiding Values, Relational Matrix (五伦, Wulun), Network Strength, 4 Mindsets, Mintzberg Mindset (all employees), Coaching Effectiveness
- 八个发展框架：8 个生活习惯、TPC 指导价值观、关系矩阵（五伦）、网络强度、4 种思维模式、明茨伯格思维模式（所有员工）、教练效能
- One Tier 3 framework: Consciousness Calibration
- 一个三级框架：意识校准
- Two derived composites: LVTS and PRCS
- 两个派生综合指标：LVTS 和 PRCS
- Track B data firewall enforced at both application and database layers
- 在应用层和数据库层强制执行的 B 轨数据防火墙
- All wave timelines and sprint assignments as specified in Sections 7 and 8
- 第 7 节和第 8 节中指定的所有波次时间表和冲刺分配

### 4.2 Out of Scope (Phase 2+)
### 4.2 范围外（第二阶段+）

The following features are explicitly not included in the MVP:
以下功能明确不包括在 MVP 中：

| Feature | Reason for deferral |
| 功能 | 延期原因 |
|---|---|
| A-28: CV connector (D2 context enrichment) | Lower RICE (6.0); Track B scope requires capacity relief |
| A-28：CV 连接器（D2 背景丰富） | 较低的 RICE（6.0）；B 轨范围需要容量释放 |
| A-29: iLevel connector (senior leader role enrichment) | Lower RICE; credential confirmation risk; Phase 2 if Sprint 8 slips |
| A-29：iLevel 连接器（高级领导者角色丰富） | 较低的 RICE；凭证确认风险；如果冲刺 8 滑点则进入第二阶段 |
| B-16: Development ROI pilot report | Requires Career Development Digital Capture (OI-07, P&O action) — not ready by MVP |
| B-16：发展 ROI 试点报告 | 需要职业发展数字捕获（OI-07，P&O 行动）— MVP 前未准备好 |
| Full MECS (Track B enriched) | Requires Track B Coaching Effectiveness data at scale |
| 完整 MECS（B 轨丰富） | 需要规模化的 B 轨教练效能数据 |
| LVTAR ≥55% | Phase 2 target, requires Track B data at scale |
| LVTAR ≥55% | 第二阶段目标，需要规模化的 B 轨数据 |
| LinkedIn Learning | Enterprise licence not confirmed; Phase 2 candidate |
| LinkedIn Learning | 企业许可证未确认；第二阶段候选 |

### 4.3 October 15 Delivery Commitment
### 4.3 10月15日交付承诺

TalentVerse goes live for all 240 OCTAVE employees on October 15, 2026. This is a firm delivery target. SMP Pillars are not included in Track B, which reduces G3 pipeline complexity. Mintzberg Mindset applies to all employees, reusing the same prompt engineering and wave structure. Neither adjustment introduces net delivery risk. The scope relief applied at the deferred features list (A-28, A-29, B-16) is sufficient to maintain the schedule.
TalentVerse 于 2026年10月15日为所有 240 名 OCTAVE 员工上线。这是一个坚定的交付目标。SMP 支柱不包括在 B 轨中，这减少了 G3 管道复杂性。明茨伯格思维模式适用于所有员工，重用相同的提示工程和波次结构。两项调整都不会引入净交付风险。在延期功能列表（A-28、A-29、B-16）应用的范围内释放足以维持时间表。

---

## 5. Data Architecture
## 5. 数据架构

### 5.1 Layer 1 — Core Signals Sources (14 sources, Track A)
### 5.1 第一层 — 核心信号源（14 个来源，A 轨）

These 14 sources feed D1-D5 and M1-M3 dimensional insights through the Medallion architecture.
这 14 个来源通过奖章架构为 D1-D5 和 M1-M3 维度洞察提供信息。

| Source | Wave | Primary Track A Signals | Integration Method |
| 来源 | 波次 | 主要 A 轨信号 | 集成方法 |
|---|---|---|---|
| BIPO HRMS | Wave 1 | Identity baseline, D1, D2 partial | REST API (OAuth2 client credentials) |
| BIPO HRMS | 波次 1 | 身份基线、D1、D2 部分 | REST API（OAuth2 客户端凭证） |
| SharePoint (MS Graph) | Wave 1 | Document infrastructure only | MS Graph API (OAuth2 Azure AD) |
| SharePoint（MS Graph） | 波次 1 | 仅文档基础设施 | MS Graph API（OAuth2 Azure AD） |
| PCG Forms | Wave 1/3 | D1-D5, M1-M3 partial (highest signal density) | MS Graph API → Azure Document Intelligence (XLSX) |
| PCG 表格 | 波次 1/3 | D1-D5、M1-M3 部分（最高信号密度） | MS Graph API → Azure 文档智能（XLSX） |
| Profit.co OKR | Wave 2 | D1, D4, D5 partial | REST API (API Key) |
| Profit.co OKR | 波次 2 | D1、D4、D5 部分 | REST API（API 密钥） |
| Profit.co ePCG | Wave 3 | D1-D5 partial, M1-M3 partial | REST API async (API Key + Username) |
| Profit.co ePCG | 波次 3 | D1-D5 部分、M1-M3 部分 | REST API 异步（API 密钥 + 用户名） |
| OCTAVE Events | Wave 3 | D2, D4 partial | MS Graph API → JSON parse (SharePoint bridge) |
| OCTAVE 活动 | 波次 3 | D2、D4 部分 | MS Graph API → JSON 解析（SharePoint 桥接） |
| Microsoft Viva Insights | Wave 4 | D3 quantitative (M365 E5 gate) | MS Graph API (OAuth2 Azure AD) |
| Microsoft Viva Insights | 波次 4 | D3 定量（M365 E5 门控） | MS Graph API（OAuth2 Azure AD） |
| 360 Feedback Survey | Wave 4 | D3, M1-M3 primary | MS Graph API → Azure Document Intelligence → NLP |
| 360 反馈调查 | 波次 4 | D3、M1-M3 主要 | MS Graph API → Azure 文档智能 → NLP |
| RMRR Reports | Wave 4 | D4 | MS Graph API → custom extractor (rmrr_report_extractor.py) |
| RMRR 报告 | 波次 4 | D4 | MS Graph API → 自定义提取器（rmrr_report_extractor.py） |
| Job Profile | Wave 4 | D1-D4 context, Mintzberg enrichment | MS Graph API → Azure Document Intelligence |
| 职位简介 | 波次 4 | D1-D4 背景、明茨伯格丰富 | MS Graph API → Azure 文档智能 |
| MT Journal | Wave 4 | D4, M1, M3 partial | MS Graph API → Azure AI Language NLP |
| MT 日志 | 波次 4 | D4、M1、M3 部分 | MS Graph API → Azure AI 语言 NLP |
| SHL OPQ32 | Wave 5 | D2-D5, M1-M3 (35 senior leaders) | MS Graph API → Azure Document Intelligence (custom extractors) |
| SHL OPQ32 | 波次 5 | D2-D5、M1-M3（35 名高级领导者） | MS Graph API → Azure 文档智能（自定义提取器） |
| SHL MQ M5 | Wave 5 | D2, M2 motivation signals | MS Graph API → Azure Document Intelligence (custom extractors) |
| SHL MQ M5 | 波次 5 | D2、M2 动机信号 | MS Graph API → Azure 文档智能（自定义提取器） |
| SHL Verify G+ | Wave 5 | D1, D5 cognitive | MS Graph API → Azure Document Intelligence (custom extractors) |
| SHL Verify G+ | 波次 5 | D1、D5 认知 | MS Graph API → Azure 文档智能（自定义提取器） |

**Key integration decisions:**
**关键集成决策：**

SharePoint is elevated to explicit P0 architectural dependency. Tenant admin consent from TPC IT is a Sprint 1 hard gate. Without it, all document-based sources are blocked.
SharePoint 提升为显式 P0 架构依赖项。来自 TPC IT 的租户管理员同意是冲刺 1 的硬门控。没有它，所有基于文档的来源都被阻止。

TalentVerse uses Microsoft Teams built-in transcription via MS Graph API for 1-on-1 transcript ingestion. This is the production path. The processed MD output files from the prototype era remain valid as the Layer 2 output schema reference.
TalentVerse 使用 MS Graph API 通过 Microsoft Teams 内置转录进行一对一转录摄取。这是生产路径。原型时代的处理后 MD 输出文件仍然有效作为第二层输出模式参考。

The Copilot ePCG DOCX file is deprecated as a data source. The underlying structured signals come directly from Profit.co ePCG (Layer 1).
Copilot ePCG DOCX 文件作为数据源已弃用。底层结构化信号直接来自 Profit.co ePCG（第一层）。

### 5.2 Layer 2 — Development Sources (5 sources, Track B)
### 5.2 第二层 — 发展来源（5 个来源，B 轨）

These 5 sources feed the Track B inference pipeline. They do not contribute directly to D1-D5 or M1-M3 dimensional insights.
这 5 个来源为 B 轨推理管道提供信息。它们不直接为 D1-D5 或 M1-M3 维度洞察做出贡献。

| Source | Wave | Primary Track B Frameworks | Integration Method |
| 来源 | 波次 | 主要 B 轨框架 | 集成方法 |
|---|---|---|---|
| 1-on-1 Check-in (Teams Transcripts) | Wave 3 | All 9 frameworks; Coaching Effectiveness | MS Graph API (Teams transcription) → Bronze → Insights Engine |
| 一对一检查（Teams 转录） | 波次 3 | 所有 9 个框架；教练效能 | MS Graph API（Teams 转录） → 青铜 → 洞察引擎 |
| OCTAVE Events (MEP, QL, CCC) | Wave 3 | 8 Living Habits, Consciousness Calibration, TPC Guiding Values | MS Graph API → JSON parse |
| OCTAVE 活动（MEP、QL、CCC） | 波次 3 | 8 个生活习惯、意识校准、TPC 指导价值观 | MS Graph API → JSON 解析 |
| Reflection Journal | Wave 4 | 8 Living Habits, 4 Mindsets, Relational Matrix, Consciousness Calibration | New capture mechanism (ADR-18) |
| 反思日志 | 波次 4 | 8 个生活习惯、4 种思维模式、关系矩阵、意识校准 | 新捕获机制（ADR-18） |
| Coaching Reports (SharePoint, nullable) | Wave 4 | 8 Living Habits, 4 Mindsets, Consciousness Calibration, Relational Matrix | MS Graph API → NLP |
| 教练报告（SharePoint，可为空） | 波次 4 | 8 个生活习惯、4 种思维模式、意识校准、关系矩阵 | MS Graph API → NLP |
| MLP/MEP Reflection Documents (SharePoint) | Wave 4 | 8 Living Habits, TPC Guiding Values, Consciousness Calibration | MS Graph API → NLP |
| MLP/MEP 反思文档（SharePoint） | 波次 4 | 8 个生活习惯、TPC 指导价值观、意识校准 | MS Graph API → NLP |

**Note on OCTAVE Events dual-layer role.** OCTAVE Events appears in both Layer 1 (Wave 3, D2 signal for Track A) and Layer 2 (Wave 3, primary signal for 8 Living Habits and Consciousness Calibration for Track B). The SharePoint-cached JSON is the production data source for MVP. If the OCTAVE App REST API is confirmed by May 2026, the API path is adopted. The JSON schema is identical regardless of retrieval path.
**关于 OCTAVE 活动双层角色的说明。** OCTAVE 活动在第一层（波次 3，A 轨的 D2 信号）和第二层（波次 3，B 轨的 8 个生活习惯和意识校准的主要信号）中都出现。SharePoint 缓存的 JSON 是 MVP 的生产数据源。如果 OCTAVE 应用 REST API 在 2026年5月前确认，则采用 API 路径。无论检索路径如何，JSON 模式都是相同的。

### 5.3 KIV Sources (Bronze only, not scored)
### 5.3 KIV 来源（仅青铜层，不评分）

| Source | Format | Reason for KIV Status |
| 来源 | 格式 | KIV 状态原因 |
|---|---|---|
| Aura Photography | PDF | Non-validated instrument; scientific basis does not meet TalentVerse evidence standards |
| 光环摄影 | PDF | 非经验证仪器；科学基础不符合 TalentVerse 证据标准 |
| TimeWaver | PDF | Non-validated instrument; bioenergetic assessment method |
| TimeWaver | PDF | 非经验证仪器；生物能量评估方法 |
| Hawkins Consciousness Scale | PDF | Non-validated calibration instrument; distinct from TalentVerse Consciousness Calibration which uses behavioral evidence |
| 霍金斯意识量表 | PDF | 非经验证校准仪器；不同于使用行为证据的 TalentVerse 意识校准 |
| 奇门遁甲 (Qi Men Dun Jia) | PDF | Metaphysical assessment; permanently excluded from inference |
| 奇门遁甲 | PDF | 形而上学评估；永久排除在推理之外 |

KIV sources are routed to /bronze/kiv/ at Bronze ingestion via the KIV Classification Filter. They are logged in the KIVLog audit table. They are never processed into Silver or Gold.
KIV 来源在青铜摄取时通过 KIV 分类过滤器路由到 /bronze/kiv/。它们在 KIVLog 审计表中记录。它们永远不会被处理成白银或黄金。

### 5.4 Wave Assignment Summary
### 5.4 波次分配摘要

| Wave | Dates | Sources Added |
| 波次 | 日期 | 添加的来源 |
|---|---|---|
| Wave 1 | Apr 6 - May 1 | BIPO HRMS, SharePoint (MS Graph), PCG Forms (schema), KIV instruments |
| 波次 1 | 4月6日 - 5月1日 | BIPO HRMS、SharePoint（MS Graph）、PCG 表格（模式）、KIV 仪器 |
| Wave 2 | May 4 - May 29 | Profit.co OKR, Profit.co ePCG, Teams transcript ingestion |
| 波次 2 | 5月4日 - 5月29日 | Profit.co OKR、Profit.co ePCG、Teams 转录摄取 |
| Wave 3 | Jun 1 - Jun 26 | PCG Forms (connector), OCTAVE Events, 1-on-1 Insights Engine |
| 波次 3 | 6月1日 - 6月26日 | PCG 表格（连接器）、OCTAVE 活动、一对一洞察引擎 |
| Wave 4 | Jun 29 - Aug 7 | 360 Feedback, RMRR, Job Profile, Viva Insights, Reflection Journal, Coaching Reports, MLP/MEP Reflections, MT Journal |
| 波次 4 | 6月29日 - 8月7日 | 360 反馈、RMRR、职位简介、Viva Insights、反思日志、教练报告、MLP/MEP 反思、MT 日志 |
| Wave 5 | Aug 10 - Sep 18 | SHL OPQ32/MQ/Verify G+ (35 senior leaders), ML model training |
| 波次 5 | 8月10日 - 9月18日 | SHL OPQ32/MQ/Verify G+（35 名高级领导者）、ML 模型训练 |
| Wave 6 | Sep 8 - Oct 15 | SHAP/LIME, EVS, LVTS, PRCS, Digital Twin UX, Calibration Tools, UAT |
| 波次 6 | 9月8日 - 10月15日 | SHAP/LIME、EVS、LVTS、PRCS、数字孪生 UX、校准工具、UAT |

---

## 6. Development Framework Layer (Track B)
## 6. 发展框架层（B 轨）

### 6.1 Architecture
### 6.1 架构

Track B infers development profiles from unstructured language. No Track B insight is calculated from a structured rating field. Every insight is AI-inferred from transcript evidence, reflection text, and programme participation records. Every inference produces evidence citations: the specific transcript moments or reflection passages that drove the insight.
B 轨从非结构化语言推断发展概况。没有 B 轨洞察是从结构化评分字段计算的。每个洞察都是从转录证据、反思文本和项目参与记录 AI 推断的。每个推断产生证据引用：推动洞察的特定转录时刻或反思段落。

### 6.2 Track B: Growth and Development — 4 Growth Dimensions
### 6.2 B 轨：成长与发展 — 4 个成长维度

Track B organises employee growth around four named dimensions grounded in TPC philosophy and the I-to-We journey. These dimensions replace the Tier 1/2/3 labels in the employee-facing product. Engineering tier activation is preserved as internal gating logic.
B 轨围绕扎根于 TPC 哲学和 I-to-We 旅程的四个命名维度组织员工成长。这些维度在面向员工的产品中取代了 Tier 1/2/3 标签。工程层级激活保留为内部门控逻辑。

| Dimension | Chinese Concept | Pinyin | Sphere | Constituent Frameworks | Activation |
| 维度 | 中文概念 | 拼音 | 领域 | 组成框架 | 激活 |
|---|---|---|---|---|---|
| G1: Self-Cultivation | 修身 | Xiushen | Individual | 8 Living Habits, TPC Guiding Values | Wave 4 (Tier 1) |
| G1: 自我修养 | 修身 | Xiushen | 个人 | 8 个生活习惯、TPC 指导价值观 | 波次 4（一级） |
| G2: Roles and Relationships | 齐家 | Qijia | Family / Team | Relational Matrix (五伦, Wulun), Network Strength | Wave 5 (Tier 2) |
| G2: 角色与关系 | 齐家 | Qijia | 家庭 / 团队 | 关系矩阵（五伦）、网络强度 | 波次 5（二级） |
| G3: Stewardship | 治国 | Zhiguo | Organisation | 4 Mindsets, Mintzberg Mindset, Coaching Effectiveness | Wave 5 (Tier 2) |
| G3: 管理 | 治国 | Zhiguo | 组织 | 4 种思维模式、明茨伯格思维模式、教练效能 | 波次 5（二级） |
| G4: Consciousness | 平天下, 天人合一 | Ping tianxia, Tian ren he yi | Universal | Consciousness Calibration | Wave 6 (Tier 3) |
| G4: 意识 | 平天下、天人合一 | Ping tianxia, Tian ren he yi | 普遍 | 意识校准 | 波次 6（三级） |

The IRTF cycle (In-Search, Research, Think-Through, Follow-Through) is the primary inquiry frame for Track B. IRTF is not a review-time tool. It runs continuously. Each IRTF cycle generates evidence that surfaces across one or more G dimensions. The AACC cycle gives direction: what an employee Co-creates in the performance track surfaces the signal that starts the next In-Search. What they Follow-Through on in growth changes how they are Aware in the next performance cycle.
IRTF 周期（内寻、探究、深思、践行）是 B 轨的主要探究框架。IRTF 不是审查时间工具。它持续运行。每个 IRTF 周期产生在一个或多个 G 维度上呈现的证据。AACC 周期提供方向：员工在绩效轨道上共创的内容呈现开始下一次内寻的信号。他们在成长中践行的事项改变了他们在下一个绩效周期中如何觉察。

Two composites sit above the four dimensions. LVTS (Longitudinal Value Trajectory) measures directional change across G1-G4 over time. PRCS (Promotional Readiness Confidence) combines Track A performance evidence with Track B growth trajectory. Both require data across multiple sessions before they are meaningful.
两个综合指标位于四个维度之上。LVTS（纵向价值轨迹）衡量 G1-G4 随时间的方向变化。PRCS（晋升准备信心）结合 A 轨绩效证据与 B 轨成长轨迹。两者都需要多会话的数据才能有意义。

Track B data is governed by a separate consent architecture and data firewall. Track B inference outputs are never included in the calibration panel data. Employees and their direct managers can see Track B data. The HRBP view of Track B is limited to G4 (Consciousness) and PRCS for succession planning purposes only.
B 轨数据由单独的同意架构和数据防火墙管理。B 轨推理输出永远不会包括在校准小组数据中。员工及其直接经理可以看到 B 轨数据。HRBP 对 B 轨的视图仅限于 G4（意识）和 PRCS，仅用于继任规划目的。

### 6.3 Constituent Framework Detail
### 6.3 组成框架详情

The nine frameworks activate progressively by wave. Each tier requires progressively more data history and carries progressively greater cultural and philosophical sensitivity.
九个框架按波次逐步激活。每个层级需要越来越多的数据历史，并承载越来越大的文化和哲学敏感性。

**G1: Self-Cultivation — Tier 1 Frameworks (Waves 3-4)**
**G1: 自我修养 — 一级框架（波次 3-4）**

Tier 1 frameworks can be inferred from a single session. They produce meaningful first readings with minimal data history.
一级框架可以从单个会话推断。它们在最小数据历史的情况下产生有意义的第一读数。

**8 Living Habits** represent TPC's cultural DNA. They describe how a person engages with daily work and relationships at a behavioral level.
**8 个生活习惯** 代表 TPC 的文化 DNA。它们描述一个人在行为层面如何参与日常工作和关系。

| Habit | What It Describes |
| 习惯 | 描述内容 |
|---|---|
| Breathwork | Breath-based regulation practices; using breathwork as a foundation for presence and focus |
| 呼吸法 | 基于呼吸的调节实践；使用呼吸法作为临在和专注的基础 |
| Eating | Intentional relationship with nourishment; treating food as a dimension of personal sustainability |
| 饮食 | 与滋养的意向关系；将食物视为个人可持续性的一个维度 |
| Sleeping | Sleep as a recovery and renewal discipline; rest as a leadership practice |
| 睡眠 | 作为恢复和更新纪律的睡眠；休息作为领导力实践 |
| Movement | Physical engagement as part of vitality and energy management |
| 运动 | 作为活力和管理能量一部分的身体参与 |
| Stillness | Creating space for reflection, silence, and inner listening |
| 静止 | 为反思、沉默和内在倾听创造空间 |
| Exploration of Life | Curiosity-driven engagement with new experiences, ideas, and growth edges |
| 探索生活 | 好奇心驱动的参与新体验、新想法和成长边缘 |
| Emotional Centredness | Capacity to remain grounded and regulated across varied emotional contexts |
| 情绪中心 | 在各种情绪情境中保持扎根和调节的能力 |
| Celebration of Life | Gratitude and appreciation as active practices; marking milestones and honoring others |
| 庆祝生活 | 感恩和欣赏作为主动实践；标记里程碑和尊重他人 |

Insights are expressed as a trajectory across sessions (Emerging, Developing, Established). Evidence accumulates across sessions. A habit with strong signal in Q3 but no signal in Q4 will have its weighting decay.
洞察以跨会话的轨迹表达（新兴、发展中、已建立）。证据跨会话积累。第三季度信号强劲但第四季度没有信号的习惯的权重将衰减。

**TPC Guiding Values** describe the organizational values that every employee is expected to embody and demonstrate.
**TPC 指导价值观** 描述每位员工都应该体现和展示的价值观。

| Value | What It Describes |
| 价值观 | 描述内容 |
|---|---|
| Brand Credibility | Consistent representation of TPC's mission and standards in all interactions |
| 品牌可信度 | 在所有互动中始终代表 TPC 的使命和标准 |
| Stewardship | Long-term responsibility for people, resources, and institutional purpose |
| 管理 | 对人员、资源和机构目的的长期责任 |
| Entrepreneurship | Initiative, innovation, and proactive contribution beyond role boundaries |
| 创业精神 | 主动性、创新和角色边界之外的主动贡献 |
| Value Add | Delivering meaningful contribution to colleagues, clients, and the broader community |
| 增值 | 为同事、客户和更广泛的社区提供有意义的贡献 |
| Culture | Active participation in and reinforcement of TPC's cultural DNA |
| 文化 | 积极参与和强化 TPC 的文化 DNA |
| Coherency | Alignment between stated commitments and observed behavior |
| 连贯性 | 陈述承诺与观察行为之间的对齐 |

Insights are inferred from how the employee describes their work and relationships across sessions, not from self-reported ratings. Wave 4 activation.
洞察是从员工如何跨会话描述他们的工作和关系推断的，而不是从自我报告的评分。波次 4 激活。

**G2: Roles and Relationships — Tier 2 Frameworks (Wave 5)**
**G2: 角色与关系 — 二级框架（波次 5）**

Tier 2 frameworks require 3-5 sessions across at least two months before stable insights are produced.
二级框架需要 3-5 个会话跨至少两个月才能产生稳定的洞察。

**Relational Matrix (五伦, Wulun)** draws from five relational orientations describing how an individual holds their key relationships at work.
**关系矩阵（五伦）** 从描述个人如何在工作中把握关键关系的五种关系取向中提取。

| Code | Orientation | What It Describes |
| 代码 | 取向 | 描述内容 |
|---|---|---|
| 对上以敬 | Respect upward | How the employee relates to those in authority: appropriate deference while maintaining integrity |
| 对上以敬 | 向上尊重 | 员工如何与权威人士建立关系：在保持完整性的同时适当的尊重 |
| 对下以慈 | Care downward | How the employee leads and cares for direct reports and junior colleagues |
| 对下以慈 | 向下关怀 | 员工如何领导和关怀直接下属和初级同事 |
| 对人以和 | Harmony with peers | How the employee maintains constructive, collaborative relationships with colleagues |
| 对人以和 | 与同伴和谐 | 员工如何与同事保持建设性、协作关系 |
| 对事以真 | Truthfulness in work | How the employee approaches responsibilities with integrity and honesty |
| 对事以真 | 工作中的真实 | 员工如何以完整性和诚实对待责任 |
| 对生命增值 | Add value to life | How the employee contributes beyond task completion to others' growth and collective meaning |
| 对生命增值 | 为生活增值 | 员工如何在任务完成之外为他人的成长和集体意义做出贡献 |

Insights are expressed on a maturity trajectory: Insufficient Data, Developing, Competent, Established, Leading. Cultural sensitivity note: employees who grew up outside frameworks emphasizing these relational orientations may initially produce lower-signal readings. This reflects data sparsity, not behavioral deficiency. Employee communication must make this explicit before Wave 5 activation.
洞察以成熟轨迹表达：数据不足、发展中、胜任、已建立、领导。文化敏感性说明：在这些关系取向不是重点的框架之外成长的员工最初可能产生较低信号的读数。这反映了数据稀疏性，而不是行为缺陷。员工沟通必须在波次 5 激活之前明确这一点。

**Network Strength.** Six dimensions: Internal Network Breadth, Internal Network Quality, External Network Breadth, External Network Quality, Cross-Functional Collaboration, Stakeholder Management. Overall label: Isolated, Developing, Connected, Influential. Primary source: Viva Insights (Layer 1) for internal network. CRM is required for external network breadth and is not available until Wave 5.
**网络强度。** 六个维度：内部网络广度、内部网络质量、外部网络广度、外部网络质量、跨职能协作、利益相关者管理。总体标签：孤立、发展中、已连接、有影响力。主要来源：Viva Insights（第一层）用于内部网络。CRM 需要用于外部网络广度，波次 5 之前不可用。

**G3: Stewardship — Tier 2 Frameworks (Wave 5)**
**G3: 管理 — 二级框架（波次 5）**

Tier 2 frameworks in G3 cover the organisational mindsets and leadership practices that shape how an employee operates as a steward of the institution.
G3 中的二级框架涵盖组织思维模式和塑造员工作为机构管理者运作方式的领导力实践。

**4 Mindsets** describe the mental models that shape how a person leads and makes decisions. Where habits describe what people do, mindsets describe the orientation from which they act.
**4 种思维模式** 描述塑造一个人如何领导和做决策的心智模式。习惯描述人们做什么的地方，思维模式描述他们行动的取向。

| Mindset | What It Describes |
| 思维模式 | 描述内容 |
|---|---|
| Management Mindset | Organising, coordinating, and executing within a role and team; disciplined stewardship of resources and time |
| 管理思维模式 | 在角色和团队中组织、协调和执行；资源和时间的纪律管理 |
| Growth and Development Mindset | Treating challenges as learning opportunities; believing capability is developed through sustained effort |
| 成长与发展思维模式 | 将挑战视为学习机会；相信能力是通过持续努力发展的 |
| Entrepreneurial and Creativity Mindset | Initiative, creative contribution, and proactive pursuit of new possibilities beyond prescribed scope |
| 创业与创造思维模式 | 主动性、创造性贡献和追求规定范围之外的新可能性的主动追求 |
| Relational Mindset | Investing in relationships as a strategic priority; seeing trust-building as inseparable from effectiveness |
| 关系思维模式 | 将关系投资作为战略优先事项；将信任建设视为与效能不可分割 |

Insights expressed as a maturity trajectory: Emerging, Developing, Proficient, Leading.
洞察以成熟轨迹表达：新兴、发展中、熟练、领导。

**Mintzberg Mindset (all employees).** Henry Mintzberg's model describes the roles and mindsets through which people act in organisations. Two complementary lenses are applied.
**明茨伯格思维模式（所有员工）。** 亨利·明茨伯格的模型描述人们在组织中通过的角色和思维模式。应用了两个互补的视角。

*10 Managerial Roles* grouped in three clusters: Interpersonal (Figurehead, Leader, Liaison), Informational (Monitor, Disseminator, Spokesperson), Decisional (Entrepreneur, Disturbance Handler, Resource Allocator, Negotiator).
*10 个管理角色* 分三个集群：人际关系（挂名首脑、领导者、联络者）、信息（监控者、传播者、发言人）、决策（创业者、 disturbance handler、资源分配者、谈判者）。

*5 Gosling-Mintzberg Mindsets*: Reflective (making sense of experience), Analytic (decomposing problems), Worldly (engaging context and nuance), Collaborative (working through others), Action (getting things done).
*5 个 Gosling-明茨伯格思维模式*：反思（理解经验）、分析（分解问题）、世故（参与背景和细微差别）、协作（通过他人工作）、行动（完成事情）。

Mintzberg Mindset applies to all employees. The 1-on-1 session transcript is the primary source: the inference engine reads behavioral signals across all five mindsets from the way the employee engages, asks questions, and describes their work — regardless of whether they hold a formal management role. Insights are expressed per role cluster and per mindset on a trajectory: Not yet visible, Emerging, Developing, Leading.
明茨伯格思维模式适用于所有员工。一对一会议转录是主要来源：推理引擎从员工参与、提问和描述工作的方式读取所有五种思维模式的行为信号——无论他们是否担任正式管理角色。洞察以每个角色集群和每个思维模式的轨迹表达：尚未可见、新兴、发展中、领导。

**Coaching Effectiveness** (Wave 4 onward, managers). Across four dimensions: Coaching (asking powerful questions, building capability), Feedback Delivery (timeliness, specificity, safety), Development Support (investment in the employee's growth agenda), Communication (clarity, presence, psychological safety). Overall insight is a weighted aggregate across sessions. Visible to the manager (self-view) and HRBP. Not visible to the manager's direct reports. Sits within G3.
**教练效能**（波次 4 起，管理者）。跨四个维度：教练（提出强有力的问题、培养能力）、反馈交付（及时性、特异性、安全性）、发展支持（对员工成长议程的投资）、沟通（清晰性、临在、心理安全）。总体洞察是跨会话的加权汇总。对经理（自我视图）和 HRBP 可见。对经理的直接下属不可见。位于 G3 内。

**G4: Consciousness — Tier 3 Frameworks (Wave 6)**
**G4: 意识 — 三级框架（波次 6）**

Tier 3 frameworks carry the highest cultural and philosophical sensitivity. They require longitudinal data and a clearly communicated consent and explanation process for employees. Minimum 5 sessions required before any insight is published to an employee profile.
三级框架承载最高的文化和哲学敏感性。它们需要纵向数据和向员工明确传达的同意和解释过程。在洞察发布到员工档案之前，至少需要 5 个会话。

**Consciousness Calibration.** Describes where a person is on the I-to-We developmental journey.
**意识校准。** 描述一个人在 I-to-We 发展旅程中的位置。

| Stage | Label | Description |
| 阶段 | 标签 | 描述 |
|---|---|---|
| Stage 0 | Insufficient Data | Not enough evidence to assign a stage |
| 阶段 0 | 数据不足 | 证据不足，无法分配阶段 |
| Stage 1 | To Me | Reactive mode; world acts on the individual |
| 阶段 1 | 对我 | 反应模式；世界对个人起作用 |
| Stage 2 | By Me | Individual achievement orientation; proactive but self-referential |
| 阶段 2 | 由我 | 个人成就导向；主动但以自我为参照 |
| Stage 3 | Through Me | Channel consciousness; contributing as a conduit for team and mission |
| 阶段 3 | 通过我 | 渠道意识；作为团队和使命的渠道做出贡献 |
| Stage 4 | As Me | Unity consciousness; organisational mission and personal identity integrated |
| 阶段 4 | 如我 | 统一意识；组织使命和个人身份整合 |
| Stage 5 | Beyond Me | Systems transformation; societal and generational impact orientation |
| 阶段 5 | 超越我 | 系统转型；社会和代际影响导向 |

Insights are expressed as a qualitative stage assignment with a confidence level (Low, Medium, High). The AI does not assign a stage it cannot support with quoted evidence. Wave 6.
洞察以带有置信度水平（低、中、高）的定性阶段分配表达。AI 不会分配它无法引用证据支持的阶段。波次 6。

### 6.4 Derived Composites (Wave 6)
### 6.4 派生综合指标（波次 6）

**LVTS — Longitudinal Value Trajectory.** Measures directional change across G1-G4 over time. Expressed on a -10 to +10 directional index where 0 is no change. Positive LVTS: consistent growth. Negative LVTS: stagnation or regression. Uses a rolling 12-month window. Minimum data requirement: 3 Track B inference events across at least two separate calendar quarters. Does not compute if all evidence falls within a single quarter.
**LVTS — 纵向价值轨迹。** 衡量 G1-G4 随时间的方向变化。以 -10 到 +10 的方向指数表达，其中 0 是没有变化。正 LVTS：持续增长。负 LVTS：停滞或倒退。使用滚动 12 个月窗口。最低数据要求：跨至少两个不同日历季度的 3 个 B 轨推理事件。如果所有证据都落在单个季度内，则不计算。

**PRCS — Promotional Readiness Confidence.** Combines Track A performance evidence with Track B growth trajectory. Track A inputs: D1-D5 dimensional insights, M1-M3 (for managers), ePCG rating trajectory, OKR achievement rates. Track B inputs: 4 Mindsets aggregate (Growth and Development Mindset and Entrepreneurial and Creativity Mindset are predictive of readiness), Consciousness Calibration stage (By Me or above is a minimum threshold; Through Me is a strong positive signal). PRCS informs calibration panel discussions but does not replace human judgment.
**PRCS — 晋升准备信心。** 结合 A 轨绩效证据与 B 轨成长轨迹。A 轨输入：D1-D5 维度洞察、M1-M3（对于管理者）、ePCG 评级轨迹、OKR 达成率。B 轨输入：4 种思维模式汇总（成长与发展思维模式和创业与创造思维模式预测准备度）、意识校准阶段（由我或以上是最低阈值；通过我是强积极信号）。PRCS 为校准小组讨论提供信息，但不取代人类判断。

### 6.5 Track B Data Firewall Access Matrix
### 6.5 B 轨数据防火墙访问矩阵

| Data type | Employee | Direct Manager | HRBP | Calibration Panel |
| 数据类型 | 员工 | 直接经理 | HRBP | 校准小组 |
|---|---|---|---|---|
| G1: 8 Living Habits insights | Visible | Visible | Visible (aggregate only) | NOT visible by default |
| G1: 8 个生活习惯洞察 | 可见 | 可见 | 可见（仅汇总） | 默认不可见 |
| G1: TPC Guiding Values insights | Visible | Visible | Visible (aggregate only) | NOT visible by default |
| G1: TPC 指导价值观洞察 | 可见 | 可见 | 可见（仅汇总） | 默认不可见 |
| G2: Relational Matrix insights | Visible | NOT visible | NOT visible by default | NOT visible |
| G2: 关系矩阵洞察 | 可见 | 不可见 | 默认不可见 | 不可见 |
| G2: Network Strength | Visible | Visible | Visible (aggregate only) | NOT visible by default |
| G2: 网络强度 | 可见 | 可见 | 可见（仅汇总） | 默认不可见 |
| G3: 4 Mindsets insights | Visible | Visible | Visible (aggregate only) | NOT visible by default |
| G3: 4 种思维模式洞察 | 可见 | 可见 | 可见（仅汇总） | 默认不可见 |
| G3: Mintzberg Mindset insights | Visible | Visible | Visible | NOT visible by default |
| G3: 明茨伯格思维模式洞察 | 可见 | 可见 | 可见 | 默认不可见 |
| G3: Coaching Effectiveness | NOT visible | Self-view only | Visible | NOT visible |
| G3: 教练效能 | 不可见 | 仅自我视图 | 可见 | 不可见 |
| G4: Consciousness Calibration | Visible | NOT visible | HRBP-only, with consent | NOT visible |
| G4: 意识校准 | 可见 | 不可见 | 仅 HRBP，需同意 | 不可见 |
| Reflection Journal content | Visible (own only) | NOT visible | NOT visible | NOT visible |
| 反思日志内容 | 可见（仅自己的） | 不可见 | 不可见 | 不可见 |
| LVTS | Visible | Visible | Visible | Visible (HRBP gatekeeps) |
| LVTS | 可见 | 可见 | 可见 | 可见（HRBP 门控） |
| PRCS | Visible | Visible | Visible | Visible (with HRBP gate) |
| PRCS | 可见 | 可见 | 可见 | 可见（有 HRBP 门控） |

The firewall is enforced at both the application layer and the database layer. Policy alone is not sufficient.
防火墙在应用层和数据库层都强制执行。仅策略是不够的。

---

## 7. Stage-Based 2026 Roadmap
## 7. 基于阶段的 2026 路线图

### Stage A — Performance Foundation + Track B Ingestion (Q2-Q3 2026, Waves 1-4)
### 阶段 A — 绩效基础 + B 轨摄取（2026年第二季度-第三季度，波次 1-4）

**Gate milestone**: August 8, 2026 (end of Sprint 9)
**门控里程碑**：2026年8月8日（冲刺 9 结束）

**What is live at Stage A gate:**
**阶段 A 门控时上线的内容：**
- All Layer 1 connectors active or in go/no-go evaluation (BIPO, OKR, ePCG, PCG Form, 360, RMRR, Job Profile)
- 所有第一层连接器活跃或处于 go/no-go 评估（BIPO、OKR、ePCG、PCG 表格、360、RMRR、职位简介）
- D1-D5 and M1-M3 each have at least one strong signal
- D1-D5 和 M1-M3 每个至少有一个强信号
- Track B Layer 2 ingestion flowing: Teams transcripts accumulating since Wave 2, OCTAVE Events live, Reflection Journal accepting entries
- B 轨第二层摄取流动：Teams 转录自波次 2 起累积，OCTAVE 活动上线，反思日志接受条目
- Track B data firewall live and tested
- B 轨数据防火墙上线并测试
- Consent Dashboard live with all 4 tiers plus Tier 2B (transcript consent)
- 同意仪表板上线，所有 4 层加上 Tier 2B（转录同意）
- Tier 1 inference running: 8 Living Habits insights for employees with 1+ processed sessions; TPC Guiding Values insights beginning to accumulate
- 一级推理运行：有 1+ 处理会话的员工的 8 个生活习惯洞察；TPC 指导价值观洞察开始累积
- Coaching Effectiveness insights live for managers (self-view only)
- 管理者教练效能洞察上线（仅自我视图）
- Manager dashboard updated with Stage A performance evidence
- 管理者仪表板更新阶段 A 绩效证据

**LVTAR target at Stage A gate**: ≥10% (Track B data beginning to accumulate)
**阶段 A 门控时的 LVTAR 目标**：≥10%（B 轨数据开始累积）

### Stage B — Dual-Track Intelligence + Go-Live (Q3-Q4 2026, Waves 5-6)
### 阶段 B — 双轨智能 + 上线（2026年第三季度-第四季度，波次 5-6）

**Gate milestone**: October 15, 2026 (end of Sprint 14)
**门控里程碑**：2026年10月15日（冲刺 14 结束）

**What is live at go-live:**
**上线时的内容：**
- Full Track A insights: 3D/5D dimensional insights, EVS (confidence-weighted composite), MECS-baseline, SHAP/LIME explainability
- 完整 A 轨洞察：3D/5D 维度洞察、EVS（置信度加权综合）、MECS-基线、SHAP/LIME 可解释性
- Full Track B inference: all nine frameworks, LVTS, PRCS
- 完整 B 轨推理：所有九个框架、LVTS、PRCS
- Employee Digital Twin with both tracks visible to the employee (self-view)
- 员工数字孪生，员工可见两轨（自我视图）
- AI-generated development insights and recommendations
- AI 生成的发展洞察和建议
- Manager team dashboard: Track A + Track B in one view
- 管理者团队仪表板：一个视图中的 A 轨 + B 轨
- HRBP dashboard: BU heatmaps, talent risk, EIDR view, Track B profile completion rates, PRCS distribution
- HRBP 仪表板：BU 热图、人才风险、EIDR 视图、B 轨概况完成率、PRCS 分布
- Calibration Support Tools: side-by-side comparison, evidence surfacing, rationale documentation
- 校准支持工具：并排比较、证据呈现、理由文档
- EIDR instrumented for first calibration cycle
- 为第一个校准周期提供 EIDR 仪器
- 4-persona UAT complete (Strategic Sarah, Mid-Level Marcus, Partner Priya, Growing Grace)
- 4 角色 UAT 完成（战略性 Sarah、中层 Marcus、合作伙伴 Priya、成长中的 Grace）
- PDPA compliance audit passed
- PDPA 合规审计通过

**LVTAR target at go-live**: ≥25%
**上线时的 LVTAR 目标**：≥25%
**EIDR target in first calibration cycle**: ≥20%
**第一次校准周期中的 EIDR 目标**：≥20%

**Wave 5 Tier 2 activation list:**
**波次 5 二级激活列表：**
- Mintzberg Mindset (all employees)
- 明茨伯格思维模式（所有员工）
- Network Strength
- 网络强度
- Coaching Effectiveness (managers)
- 教练效能（管理者）

### Key Milestones Table
### 关键里程碑表

| Milestone | Date | Deliverable |
| 里程碑 | 日期 | 交付物 |
|---|---|---|
| Wave 1 complete | May 1, 2026 | Azure infra, BIPO, SharePoint, RBAC, identity service live |
| 波次 1 完成 | 2026年5月1日 | Azure 基础设施、BIPO、SharePoint、RBAC、身份服务上线 |
| Wave 2 complete | May 29, 2026 | OKR, ePCG connectors live; Teams transcript ingestion accumulating |
| 波次 2 完成 | 2026年5月29日 | OKR、ePCG 连接器上线；Teams 转录摄取累积 |
| W13 Leadership Demo | May 29, 2026 (Week 8) | 3-source Impact Rating from BIPO + OKR + Job Profile demonstrated |
| W13 领导力演示 | 2026年5月29日（第 8 周） | 展示来自 BIPO + OKR + 职位简介的 3 源影响评级 |
| Wave 3 complete | Jun 26, 2026 | PCG Form, 1-on-1 Insights Engine, OCTAVE Events, manager dashboard skeleton |
| 波次 3 完成 | 2026年6月26日 | PCG 表格、一对一洞察引擎、OCTAVE 活动、管理者仪表板骨架 |
| Stage A gate | Aug 8, 2026 | All Layer 1 connectors active; Track B ingestion flowing; Tier 1 inference live |
| 阶段 A 门控 | 2026年8月8日 | 所有第一层连接器活跃；B 轨摄取流动；一级推理上线 |
| Wave 5 complete | Sep 18, 2026 | SHL connector, ML trained and bias-audited, Tier 2 inference live |
| 波次 5 完成 | 2026年9月18日 | SHL 连接器、ML 训练和偏见审计、二级推理上线 |
| Go-live | Oct 15, 2026 | Full dual-track MVP for 240 employees |
| 上线 | 2026年10月15日 | 240 名员工的完整双轨 MVP |

### Phase 2+ Horizon (H1 2027 Onward)
### 第二阶段+ 视野（2027年上半年起）

Phase 2 begins after October 15, 2026. Prerequisites that must begin by Q3 2026: Career Development Digital Capture (OI-07: P&O + Data Engineering parallel workstream, not a product team task). Structured reflection templates must be standardised before MLP/MEP Reflection extraction is reliable.
第二阶段在 2026年10月15日之后开始。必须在 2026年第三季度开始的先决条件：职业发展数字捕获（OI-07：P&O + 数据工程并行工作流，不是产品团队任务）。结构化反思模板必须在 MLP/MEP 反思提取可靠之前标准化。

**Phase 2 (H1 2027)**: Full MECS (Track B enriched), Development ROI pilot (cohort: programme entry to 6-month Track B signal delta), CV connector, iLevel connector. LVTAR target: ≥55%.
**第二阶段（2027年上半年）**：完整 MECS（B 轨丰富）、发展 ROI 试点（队列：项目进入到 6 个月 B 轨信号增量）、CV 连接器、iLevel 连接器。LVTAR 目标：≥55%。

**Phase 3 (H2 2027-2028)**: Onboarding baseline and origin context. SHL direct API replacing PDF extraction. LVTAR target: ≥75%.
**第三阶段（2027年下半年-2028年）**：入职基线和起源背景。SHL 直接 API 替代 PDF 提取。LVTAR 目标：≥75%。

**Phase 4 (2028)**: Full 4-signal MECS, FREW (Flight Risk Early Warning). LVTAR target: ≥90%.
**第四阶段（2028年）**：完整 4 信号 MECS、FREW（离职风险预警）。LVTAR 目标：≥90%。

**Phase 5 (2028-2029)**: Full Lifecycle OS. LVTAR target: ≥95%. External offering path opens.
**第五阶段（2028-2029年）**：完整生命周期操作系统。LVTAR 目标：≥95%。外部产品路径开放。

---

## 8. Sprint Plan Summary
## 8. 冲刺计划摘要

### 8.1 Epic Overview
### 8.1 史诗概述

Six epics span 14 sprints (2-week cadence, April 6 to October 15, 2026; Sprint 14 is a 10-day sprint: Oct 6-15).
六个史诗跨越 14 个冲刺（2 周节奏，4月6日至10月15日；冲刺 14 是 10 天冲刺：10月6-15日）。

| Epic ID | Epic Name | Sprint Span | Track | Goal |
| 史诗 ID | 史诗名称 | 冲刺跨度 | 轨道 | 目标 |
|---|---|---|---|---|
| EP-01 | Data Foundation and Identity | Sprints 1-4 (Waves 1-2) | Both | All infrastructure, identity, and Layer 1 API connectors live; Teams transcript ingestion pipeline in place; W13 Leadership Demo delivered |
| EP-01 | 数据基础和身份 | 冲刺 1-4（波次 1-2） | 两轨 | 所有基础设施、身份和第一层 API 连接器上线；Teams 转录摄取管道到位；W13 领导力演示交付 |
| EP-02 | Track A Insights Engine | Sprints 5-14 (Waves 3-6) | A | All Layer 1 connectors live; 3D/5D dimensional insights with confidence weights; SHAP/LIME explainability; EVS and MECS-baseline computed for all eligible employees |
| EP-02 | A 轨洞察引擎 | 冲刺 5-14（波次 3-6） | A | 所有第一层连接器上线；3D/5D 维度洞察带置信度权重；SHAP/LIME 可解释性；所有合格员工的 EVS 和 MECS-基线计算 |
| EP-03 | Track B Framework Pipeline | Sprints 5-14 (Waves 3-6) | B | All 9 development frameworks generating insights; LVTS and PRCS computed; Track B data firewall enforced |
| EP-03 | B 轨框架管道 | 冲刺 5-14（波次 3-6） | B | 所有 9 个发展框架生成洞察；LVTS 和 PRCS 计算；B 轨数据防火墙强制执行 |
| EP-04 | Employee Experience Layer | Sprints 7-14 (Waves 4-6) | Both | Every employee can see their own Digital Twin (Track A + Track B), manage consent, and act on AI-generated development recommendations |
| EP-04 | 员工体验层 | 冲刺 7-14（波次 4-6） | 两轨 | 每位员工可以看到自己的数字孪生（A 轨 + B 轨）、管理同意并对 AI 生成的发展建议采取行动 |
| EP-05 | Manager Intelligence Layer | Sprints 5-14 (Waves 3-6) | Both | Managers have a real-time team dashboard combining Track A evidence with Track B development insights and Coaching Effectiveness |
| EP-05 | 管理者智能层 | 冲刺 5-14（波次 3-6） | 两轨 | 管理者拥有结合 A 轨证据和 B 轨发展洞察以及教练效能的实时团队仪表板 |
| EP-06 | P&O Decision Support and Go-Live | Sprints 13-14 (Wave 6) | Both | HRBP and executive dashboards calibration-ready; EIDR instrumented; UAT and PDPA audit complete; platform live October 15 |
| EP-06 | P&O 决策支持和上线 | 冲刺 13-14（波次 6） | 两轨 | HRBP 和执行仪表板校准就绪；EIDR 仪器化；UAT 和 PDPA 审计完成；平台于 10月15日上线 |

### 8.2 Sprint Assignment Summary
### 8.2 冲刺分配摘要

| Sprint | Wave | Key Deliverables |
| 冲刺 | 波次 | 关键交付物 |
|---|---|---|
| Sprint 1 | Wave 1 | Azure infra, PDPA foundation, RBAC, EmployeeIdentityMap, BIPO connector, SharePoint connector |
| 冲刺 1 | 波次 1 | Azure 基础设施、PDPA 基础、RBAC、EmployeeIdentityMap、BIPO 连接器、SharePoint 连接器 |
| Sprint 2 | Wave 1 | Identity gate (100-employee sample), LVTAR instrumentation, tenant admin consent confirmed |
| 冲刺 2 | 波次 1 | 身份门控（100 名员工样本）、LVTAR 仪器化、租户管理员同意确认 |
| Sprint 3 | Wave 2 | Profit.co OKR connector, Profit.co ePCG connector, Teams transcript ingestion pipeline, Tier 2B consent flow |
| 冲刺 3 | 波次 2 | Profit.co OKR 连接器、Profit.co ePCG 连接器、Teams 转录摄取管道、Tier 2B 同意流 |
| Sprint 4 | Wave 2 | W13 Leadership Demo, OI-02 resolved, ADR-19 LLM model selection started |
| 冲刺 4 | 波次 2 | W13 领导力演示、OI-02 解决、ADR-19 LLM 模型选择开始 |
| Sprint 5 | Wave 3 | PCG Form connector, 1-on-1 Insights Engine v1, OCTAVE Events connector, manager dashboard skeleton |
| 冲刺 5 | 波次 3 | PCG 表格连接器、一对一洞察引擎 v1、OCTAVE 活动连接器、管理者仪表板骨架 |
| Sprint 6 | Wave 3 | ADR-19 finalised, OI-04 (8 Living Habits sign-off), Wave 4 P&O communication approved |
| 冲刺 6 | 波次 3 | ADR-19 最终确定、OI-04（8 个生活习惯签署）、波次 4 P&O 沟通批准 |
| Sprint 7 | Wave 4 | 360 Feedback connector, RMRR connector, Track B data firewall, Consent Dashboard, ADR-20 |
| 冲刺 7 | 波次 4 | 360 反馈连接器、RMRR 连接器、B 轨数据防火墙、同意仪表板、ADR-20 |
| Sprint 8 | Wave 4 | Job Profile connector, Viva Insights connector, Reflection Journal, Coaching Report connector, MLP/MEP Reflection connector |
| 冲刺 8 | 波次 4 | 职位简介连接器、Viva Insights 连接器、反思日志、教练报告连接器、MLP/MEP 反思连接器 |
| Sprint 9 | Wave 4 | Tier 1 inference (8 Living Habits + TPC Guiding Values), Coaching Effectiveness, manager dashboard Stage A update, **Stage A gate** |
| 冲刺 9 | 波次 4 | 一级推理（8 个生活习惯 + TPC 指导价值观）、教练效能、管理者仪表板阶段 A 更新、**阶段 A 门控** |
| Sprint 10 | Wave 5 | SHL Psychometrics connector, ML model training begins, P&O Wave 5 gate |
| 冲刺 10 | 波次 5 | SHL 心理测评连接器、ML 模型训练开始、P&O 波次 5 门控 |
| Sprint 11 | Wave 5 | ML training complete + bias audit, 3D/5D Insights Engine, Tier 2 inference (Mintzberg Mindset all employees + Network Strength) |
| 冲刺 11 | 波次 5 | ML 训练完成 + 偏见审计、3D/5D 洞察引擎、二级推理（明茨伯格思维模式所有员工 + 网络强度） |
| Sprint 12 | Wave 5 | P&O sign-off workflow, ADR-21 Relational Matrix framing approved, ADR-22 LVTS formula confirmed |
| 冲刺 12 | 波次 5 | P&O 签署工作流、ADR-21 关系矩阵框架批准、ADR-22 LVTS 公式确认 |
| Sprint 13 | Wave 6 | SHAP/LIME explainability, EVS computed, Tier 3 inference (P&O gate required), Employee Digital Twin Track A self-view, LVTS, PRCS, Track B self-view, AI development insights, MECS-baseline, HRBP dashboard, Calibration Support Tools, EIDR instrumentation, UAT begins |
| 冲刺 13 | 波次 6 | SHAP/LIME 可解释性、EVS 计算、三级推理（需 P&O 门控）、员工数字孪生 A 轨自我视图、LVTS、PRCS、B 轨自我视图、AI 发展洞察、MECS-基线、HRBP 仪表板、校准支持工具、EIDR 仪器化、UAT 开始 |
| Sprint 14 | Wave 6 (Oct 6-15, 10 days) | UAT completion, PDPA audit, production hardening, full manager dashboard, Tier 3 self-view, **Go-Live October 15** |
| 冲刺 14 | 波次 6（10月6-15日，10 天） | UAT 完成、PDPA 审计、生产加固、完整管理者仪表板、三级自我视图、**10月15日上线** |

## 14. What Success Looks Like: October 15, 2026
## 14. 成功标准：2026年10月15日

On October 15, 2026, TalentVerse goes live for all 240 OCTAVE employees. Here is what each persona experiences.
2026年10月15日，TalentVerse 为所有 240 名 OCTAVE 员工上线。以下是每个角色体验到的内容。

### What Employees (Growing Grace) Can See
### 员工（成长中的 Grace）可以看到什么

Grace opens her Employee Digital Twin for the first time. She sees two sections.
Grace 首次打开她的员工数字孪生。她看到两个部分。

The Track A section shows her performance evidence: dimensional insights across D1-D5, a confidence level (High, Medium, or Low depending on how many sources have contributed), and an Evidence Trail showing exactly which data points drove each insight. Below each insight is a SHAP/LIME attribution: "This insight was primarily driven by PCG Form accountability sub-signals (42%) and OKR completion discipline (31%)." She can see where her evidence is strong and where it is thin. She can flag an insight she believes is inaccurate, which triggers an HRBP review.
A 轨部分显示她的绩效证据：D1-D5 的维度洞察、置信度水平（高、中或低，取决于有多少来源贡献）以及显示哪些数据点推动每个洞察的证据轨迹。每个洞察下方是 SHAP/LIME 归因："此洞察主要由 PCG 表格问责子信号（42%）和 OKR 完成纪律（31%）推动。"她可以看到她的证据在哪里强劲，在哪里薄弱。她可以对认为不准确的洞察提出标记，这会触发 HRBP 审查。

The Track B section shows her development story. She sees her 8 Living Habits as a trajectory: which habits are Established (consistent signal across sessions), which are Developing, and how the pattern has shifted since her first 1-on-1 session was processed. She sees her G3 Mindsets panel and an overall maturity label. She reads an AI-generated development narrative: three to four sentences that synthesise her growth story from the evidence, with citations to specific sessions. If her Tier 3 insights (Consciousness Calibration) are ready, she can view them in a private section of her profile visible only to her.
B 轨部分显示她的发展故事。她看到她的 8 个生活习惯作为轨迹：哪些习惯已建立（跨会话的一致信号）、哪些正在发展，以及自她第一次一对一会议处理以来模式如何转变。她看到她的 G3 思维模式面板和总体成熟标签。她阅读 AI 生成的发展叙述：从证据综合她的成长故事的三到四句话，并引用特定会话。如果她的三级洞察（意识校准）已准备好，她可以在仅对她可见的个人资料私人部分查看它们。

She also sees her consent dashboard: which data types are active, what each tier covers, and a single button to withdraw any tier. She chooses to keep everything active.
她还看到她的同意仪表板：哪些数据类型活跃、每个层级涵盖什么，以及撤回任何层级的单一按钮。她选择保持一切活跃。

### What Managers (Mid-Level Marcus) Can Do
### 管理者（中层 Marcus）可以做什么

Marcus opens his team dashboard before the calibration preparation session. He sees his full team with Track A insights: each team member's EVS, dimensional breakdown, confidence level, and a flag for any insight that is Low confidence due to insufficient source coverage. He can click into any team member to see the detailed evidence trail.
Marcus 在校准准备会议之前打开他的团队仪表板。他看到他完整的团队与 A 轨洞察：每位团队成员的 EVS、维度细分、置信度水平，以及由于源覆盖不足而导致洞察置信度低的任何标记。他可以点击进入任何团队成员以查看详细的证据轨迹。

Alongside each team member's performance view, he sees a Track B development snapshot: the most recent 8 Living Habits trajectory, the G3 Mindsets panel, and the top development insight from the last processed 1-on-1 session. He can use this to continue a coaching conversation across sessions rather than starting cold.
每位团队成员的绩效视图旁边，他看到一个 B 轨发展快照：最近的 8 个生活习惯轨迹、G3 思维模式面板，以及上次处理的一对一会议中的顶级发展洞察。他可以用这个来跨会话继续教练对话，而不是冷启动。

He also sees his own Coaching Effectiveness aggregate: how his coaching has trended across his last eight sessions, which dimensions are strongest, and one development priority identified by the system. This view is visible only to him.
他还看到他自己的教练效能汇总：他的教练在过去八次会议中的趋势、哪些维度最强，以及系统识别的一个发展优先事项。此视图仅对他可见。

When he walks into the calibration preparation meeting, he has evidence for each team member from at least three source systems. He can cite the EVS insight, show the dimensional breakdown, and point to specific sources. His HRBP partner can see the same view.
当他走进校准准备会议时，他为每位团队成员提供了至少三个源系统的证据。他可以引用 EVS 洞察、展示维度细分，并指向特定来源。他的 HRBP 合作伙伴可以看到相同的视图。

### What HRBPs (Partner Priya) Have Access To
### HRBP（合作伙伴 Priya）可以访问什么

Priya opens the HRBP dashboard before talent review. She sees BU-level EVS heatmaps: which business units have high-confidence evidence for most employees, which have data gaps. She sees talent risk flags: employees whose EVS has declined significantly or whose LVTAR suggests stagnating development.
Priya 在人才审查前打开 HRBP 仪表板。她看到 BU 级 EVS 热图：哪些业务部门对大多数员工有高置信度证据，哪些有数据差距。她看到人才风险标志：EVS 显著下降或 LVTAR 建议发展停滞的员工。

She sees Track B profile completion rates across the employee population: what percentage of employees have Tier 1 inference active, and the PRCS distribution for employees who have met the data requirements. She can filter to see employees with PRCS insights, which gives her an evidence-backed starting point for succession planning conversations.
她看到员工群体中 B 轨概况完成率：有多少百分比的员工有一级推理活跃，以及已达到数据要求的员工的 PRCS 分布。她可以过滤查看有 PRCS 洞察的员工，这为她提供了继任规划对话的证据支持起点。

She arrives at talent review with the evidence assembled. The 60%+ of her working time previously spent on manual data aggregation from six systems has been replaced by a single dashboard that she can interrogate in minutes.
她带着组装的证据到达人才审查。她以前从六个系统手动聚合数据花费的 60%+ 工作时间已被她可以在几分钟内查询的单一仪表板取代。

### What Calibration Panels See
### 校准小组看到什么

Calibration panelists see Track A evidence by default. They see side-by-side EVS comparisons, dimensional breakdowns, and evidence trails for each employee being discussed. They have a rationale documentation field: when a calibration decision is made, the decision-maker records whether TalentVerse evidence from two or more sources was cited. This is the EIDR instrumentation.
校准小组成员默认看到 A 轨证据。他们看到并排的 EVS 比较、维度细分，以及正在讨论的每位员工的证据轨迹。他们有一个理由文档字段：当做校准决定时，决策者记录是否引用了来自两个或更多来源的 TalentVerse 证据。这是 EIDR 仪器化。

Track B data is not surfaced to calibration panels by default. LVTS and PRCS can be visible if the HRBP activates access. Relational Matrix, Consciousness Calibration, and Coaching Effectiveness data are not visible to calibration panels under any conditions at MVP.
B 轨数据默认不呈现给校准小组。如果 HRBP 激活访问，可以看到 LVTS 和 PRCS。关系矩阵、意识校准和教练效能数据在 MVP 下任何条件下对校准小组都不可见。

### EIDR and LVTAR Targets Met
### EIDR 和 LVTAR 目标达成

At the first calibration cycle (likely Q1 2027, using data from the October 15 go-live system), EIDR reaches ≥20%. At least one in five calibration decisions cites TalentVerse evidence from two or more source systems in the written justification. This is the first time OCTAVE can demonstrate that talent decisions are grounded in multi-source evidence.
在第一次校准周期（可能在 2027年第一季度，使用 10月15日上线系统的数据）时，EIDR 达到 ≥20%。五分之一的校准决定在书面理由中引用了来自两个或更多源系统的 TalentVerse 证据。这是 OCTAVE 第一次可以证明人才决策基于多源证据。

LVTAR reaches ≥25% by go-live. At least one in four active Employee Digital Twins is drawing intelligence signals from two or more distinct lifecycle stages within the past 365 days: Track A performance evidence and Track B development signals together.
LVTAR 到上线时达到 ≥25%。至少四分之一的活跃员工数字孪生在过去 365 天内从两个或更多不同的生命周期阶段提取智能信号：A 轨绩效证据和 B 轨发展信号一起。

These targets are not aspirational. They are the definition of MVP success.
这些目标不是理想化的。它们是 MVP 成功的定义。

---

## 15. Phase 2 and Beyond
## 15. 第二阶段及以后

### 15.1 Phase 2 Prerequisites (H1 2027)
### 15.1 第二阶段先决条件（2027年上半年）

Phase 2 cannot begin without two actions that must start in Q3 2026. Both are P&O and Data Engineering responsibilities, not product team tasks.
没有两项必须在 2026年第三季度开始的行动，第二阶段无法开始。两项都是 P&O 和数据工程责任，不是产品团队任务。

**Career Development Digital Capture (OI-07).** Phase 2 capabilities (full MECS, Development ROI pilot, MLP/MEP Reflection extraction at scale) depend on structured reflection records from leadership programmes. Currently, reflections vary in format and depth across programmes. Standardised reflection templates must be adopted and structured capture must begin by Q3 2026 for Phase 2 extraction to be reliable by H1 2027. This is a named P&O action. The product team will surface it at every sprint review until it is confirmed.
**职业发展数字捕获（OI-07）。** 第二阶段功能（完整 MECS、发展 ROI 试点、MLP/MEP 反思大规模提取）依赖于领导力项目的结构化反思记录。目前，反思在项目中格式和深度各异。标准化反思模板必须在 2026年第三季度前采用，结构化捕获必须开始，以便第二阶段提取在 2027年上半年可靠。这是一项命名的 P&O 行动。产品团队将在每次冲刺审查中呈现它，直到确认。

**MLP/MEP Structured Records Pipeline.** Linked to the above. Data Engineering must design the ingestion schema for structured reflection documents before Phase 2 build begins. Design work should start Q3 2026.
**MLP/MEP 结构化记录管道。** 与上述相关。数据工程必须在第二阶段构建开始前设计结构化反思文档的摄取模式。设计工作应在 2026年第三季度开始。

### 15.2 Phase 2 Capabilities (H1 2027)
### 15.2 第二阶段功能（2027年上半年）

| Capability | Dependency |
| 功能 | 依赖 |
|---|---|
| Full MECS (Track B enriched) | Track B Coaching Effectiveness data at scale (6+ months of post-MVP data) |
| 完整 MECS（B 轨丰富） | 规模化的 B 轨教练效能数据（MVP 后 6+ 个月的数据） |
| Development ROI pilot report | Career Development Digital Capture (OI-07) + cohort selection (OI-06) |
| 发展 ROI 试点报告 | 职业发展数字捕获（OI-07）+ 队列选择（OI-06） |
| CV connector (A-28) | No blockers; deferred from MVP for capacity |
| CV 连接器（A-28） | 无阻碍；因容量从 MVP 延期 |
| iLevel connector (A-29) | REST API credentials confirmed |
| iLevel 连接器（A-29） | REST API 凭证确认 |
| LVTAR ≥55% | Track B data at scale |
| LVTAR ≥55% | 规模化的 B 轨数据 |

The Development ROI pilot tracks at least one cohort from programme entry (MEP or QL) to a 6-month Track B signal delta. This is the first time OCTAVE will have evidence for whether L&D investment produces measurable behavioral change.
发展 ROI 试点跟踪至少一个队列从项目进入（MEP 或 QL）到 6 个月 B 轨信号增量。这是 OCTAVE 第一次将有证据表明 L&D 投资是否产生可衡量的行为改变。

### 15.3 Phase 3 (H2 2027-2028)
### 15.3 第三阶段（2027年下半年-2028年）

Onboarding baseline for origin context. SHL direct API replacing PDF extraction for new assessments. Full MBTI and Enneagram integration. LVTAR ≥75%.
入职基线用于起源背景。SHL 直接 API 替代新评估的 PDF 提取。完整的 MBTI 和九型人格集成。LVTAR ≥75%。

### 15.4 Phase 4 (2028)
### 15.4 第四阶段（2028年）

Full 4-signal MECS. FREW (Flight Risk Early Warning) — subject to P&O governance decision on which signals are ethically appropriate for flight risk (OPEN-006 from v1). LVTAR ≥90%.
完整 4 信号 MECS。FREW（离职风险预警）— 取决于 P&O 治理决定哪些信号对离职风险在伦理上是适当的（v1 中的 OPEN-006）。LVTAR ≥90%。

### 15.5 Phase 5 (2028-2029)
### 15.5 第五阶段（2028-2029年）

Full Lifecycle OS. LVTAR ≥95%. External offering path opens.
完整生命周期操作系统。LVTAR ≥95%。外部产品路径开放。

---

*Document owner: TalentVerse PM*
*文档所有者：TalentVerse PM*
*Version: 3.0*
*版本：3.0*
*Date: April 2026*
*日期：2026年4月*

*All cost figures marked as "internal estimate" should be validated against a formal Azure Cost Management review and engineering team cost model before any board presentation. The 17 [CITATION NEEDED] items in Proposal v1 Appendix C must be resolved before external use of any market evidence claims.*
*标记为"内部估计"的所有成本数字应在任何董事会演示之前根据正式的 Azure 成本管理审查和工程团队成本模型进行验证。提案 v1 附录 C 中的 17 个[需要引用]项目必须在任何市场证据声明的外部使用之前解决。*