An open framework for turning business workflows into agent collaboration, knowledge systems, automation tools, and verification governance.
一个帮助企业将业务流程转化为智能体协作、知识库、自动化工具和验证治理机制的开源实施框架。
Turn real business workflows into verifiable AI operating systems.
把真实业务流程升级为可验证的 AI 工作体系。
Many companies are experimenting with AI, but most efforts stop at demos, chatbots, isolated tools, or one-off automations.
The real challenge is not "which model should we use?" It is:
- Which business workflows should AI enter?
- Which parts should be handled by humans, agents, tools, or systems?
- How do we connect knowledge, data, permissions, and automation?
- How do we verify AI output before it affects real work?
- How do we turn one-time experiments into durable organizational capability?
中文：
很多企业已经开始尝试 AI，但多数还停留在 demo、聊天机器人、单点工具或一次性自动化。
真正的问题不是“用哪个模型”，而是：
- AI 应该进入哪些业务流程？
- 哪些环节由人负责，哪些由智能体、工具和系统负责？
- 知识库、数据、权限、自动化如何接起来？
- AI 产出如何验证，怎么防止假结果和半成品？
- 如何把一次性尝试沉淀成组织级能力？
Enterprise AI is moving from conversation to execution. Agentic systems can plan tasks, use tools, call APIs, generate reports, and affect real operations.
That makes implementation harder. The important questions now are governance, accountability, data access, human checkpoints, auditability, and repeatable operating models.
This repository focuses on that missing implementation layer.
中文：
2026 年企业 AI 正在从“对话工具”走向“执行系统”。Agent 可以规划任务、调用工具、访问接口、生成报告，并影响真实业务。
所以真正重要的问题变成了：治理、责任、数据权限、人工审核、审计留痕，以及可复用的运行模式。
这个仓库关注的就是这层缺失的落地方法。
Business Workflow
+
Agent Collaboration
+
Knowledge System
+
Automation Tools
+
Verification Governance
=
Sustainable AI Productivity System
中文：
业务流程
+
智能体协作
+
知识库系统
+
自动化工具
+
验证治理机制
=
可持续运行的 AI 生产力系统
| Module | What it does |
|---|---|
| AI Readiness Assessment | Identify where AI can realistically create value |
| Workflow-to-Agent Mapping | Convert business processes into agent roles, tools, and human checkpoints |
| Knowledge Integration | Connect documents, data sources, policies, and permissions |
| Automation Design | Define tasks, APIs, reports, notifications, and handoffs |
| Verification Governance | Define acceptance criteria, evidence, rollback, audit, and risk gates |
| Adoption & Training | Help teams actually use the system in daily work |
中文：
| 模块 | 作用 |
|---|---|
| AI 落地成熟度评估 | 判断哪些场景适合 AI，哪些不适合 |
| 流程到智能体映射 | 把业务流程拆成 Agent、工具和人工审核节点 |
| 知识库与数据接入 | 接入文档、数据源、政策、权限和版本 |
| 自动化设计 | 定义任务、接口、报告、通知和交接 |
| 验证治理 | 定义验收、证据、回滚、审计和风险闸门 |
| 培训与组织导入 | 帮团队真正把 AI 用进日常流程 |
docs/ Core framework documents
templates/ Assessment, workflow, governance, and handoff templates
playbooks/ Step-by-step implementation playbooks
assets/ Diagrams and visual assets
- Read AI Implementation Principles
- Run the AI Readiness Assessment
- Map one workflow with Workflow-to-Agent Canvas
- Define quality gates with Verification Governance Checklist
- Use Pilot Playbook to plan a first 90-day rollout
- AI Delivery Warden: local quality gate for AI-generated deliveries
- HiMiao Hong Kong Insurance Audit Platform: domain-specific audit system example
This framework is independent, but it is aligned with the broader 2026 industry direction:
- Singapore's Model AI Governance Framework for Agentic AI emphasizes upfront risk bounding, meaningful human accountability, technical controls, and end-user responsibility.
- IBM's agentic AI governance writing highlights the execution gap: agents move AI from insight to action, which requires stronger accountability and controls.
- COMPEL and similar transformation frameworks show that enterprises need operating models, measurable outcomes, and continuous improvement, not just tools.
中文：
这套框架是独立设计的，但方向上与 2026 年主流趋势一致：
- 新加坡 Agentic AI 治理框架强调风险边界、人类负责、技术控制和终端责任。
- IBM 关于 Agentic AI 治理的内容强调：Agent 让 AI 从“提供洞察”进入“执行动作”，因此必须有更强的责任和控制机制。
- COMPEL 等企业 AI 转型框架说明：企业需要运行模型、可度量结果和持续改进，而不是只堆工具。
This framework does not replace your model provider, agent platform, or internal systems.
It provides the missing implementation layer:
business workflow -> AI operating model -> governance -> evidence -> adoption
中文：
它不替代模型、Agent 平台或企业内部系统。
它提供的是中间缺失的落地层：
业务流程 -> AI 工作模式 -> 治理机制 -> 验证证据 -> 组织导入
- More industry templates
- Governance checklist expansion
- Agent role catalog
- Example implementation cases
- Integration with AI Delivery Warden
- Bilingual presentation deck
MIT