本页以 FDE FAN 的第一视角讲解知识：先讲概念，再讲方法、图表、训练任务。底部保留参考来源，方便继续阅读。 This page teaches in FDE FAN's own voice: concepts first, then methods, visuals, and practice tasks. References are kept at the bottom for further reading.
知识结构地图Knowledge Structure Map
| 知识层级Knowledge Layer | 我们要掌握什么What We Need to Master | 训练产物Training Artifact |
|---|---|---|
| 核心论点Core thesis | FDE 式交付可以从 SIer/UX 工作坊中获得方法：先发现价值，再制作产物，再用产物提示 AI。FDE-style delivery can borrow from SIer and UX workshops: discover value, create artifacts, then prompt AI with artifacts. | 把工作坊纳入课程。Bring workshops into our curriculum. |
| 价值发现Value discovery | 客户不一定能直接说清需求，需要通过访谈、旅程、截图、现有流程和指标逐步逼近。Customers may not state requirements clearly; interviews, journeys, screenshots, existing workflows, and metrics help approach the real need. | 训练提问与观察。Train questioning and observation. |
| NSM 思维NSM thinking | 用一个核心指标统一业务目标、体验设计和 AI 方案。A single core metric aligns business goals, experience design, and AI solutions. | 让项目有可观察成功标准。Give every project an observable success criterion. |
| 产物驱动提示Artifact-driven prompting | 截图、流程图、表格、旧系统页面等，比抽象文字更能提供高质量上下文。Screenshots, flow maps, tables, and old system pages provide stronger context than abstract text. | 训练多模态上下文整理。Train multimodal context preparation. |
核心观点Key Takeaways
工作坊要产出可给 AI 使用的材料：截图、流程、字段、角色、边界和验收口径。Workshops should produce AI-usable artifacts: screenshots, flows, fields, roles, boundaries, and acceptance criteria.
NSM 能帮助 FDE 把“想要 AI”翻译成一个可观察的业务变化。NSM helps FDEs translate 'we want AI' into an observable business change.
截图和已有产物是高价值上下文，比抽象描述更容易让模型生成可验收方案。Screenshots and existing artifacts are high-value context; they often produce more testable outputs than abstract descriptions.
工作坊到交付物From Workshop to Deliverables
| 层级Layer | 含义Meaning | FDE 产物FDE Artifact |
|---|---|---|
| 发现Discover | 用户、场景、阻塞、价值假设Users, scenarios, blockers, value hypotheses | 访谈表 / 旅程图Interview brief / journey map |
| 定义Define | NSM、边界、字段、验收标准NSM, boundaries, fields, acceptance checks | 需求卡 / Prompt 模板Requirement card / prompt template |
| 交付Deliver | Demo、Agent 流程、测试和复盘Demo, agent flow, tests, retrospective | 部署链接 / README / 演示脚本Deployment link / README / demo script |
深度讲义Deep Study Notes
1. 工作坊是 FDE 的前置工程1. Workshops are pre-engineering for FDE
我们要把 FDE 的工程交付和 UX/SIer 的发现过程连接起来。很多企业 AI 项目失败不是因为模型不够强，而是因为前期没有把用户、流程、指标、限制和真实产物整理清楚。工作坊不是软性环节，而是让后续 Prompt、Demo 和 Agent 能准确落地的前置工程。We connect FDE engineering delivery with UX/SIer discovery. Many enterprise AI projects fail not because models are weak, but because users, workflows, metrics, constraints, and existing artifacts were not clarified early. Workshops are not soft activities; they are pre-engineering that enables prompts, demos, and agents to land accurately.
2. NSM 把“想要 AI”变成可验证目标2. NSM turns 'we want AI' into a verifiable objective
客户常说想要自动化、想要增长、想要客服更聪明，但这些表达太宽。NSM 的作用是把宽泛愿望收敛成一个关键变化：响应时间、转化率、人工处理量、线索质量、内容产出速度、复购率或错误率。FDE 不一定负责最终商业结果，但必须帮助客户定义可观察的试点指标。Customers often say they want automation, growth, or smarter support, but these statements are broad. NSM narrows the wish into a key change: response time, conversion rate, manual workload, lead quality, content production speed, repeat purchase, or error rate. FDEs may not own final business outcomes, but they must help define observable pilot metrics.
3. 用产物提示 AI，而不是只写抽象 Prompt3. Prompt AI with artifacts, not only abstract instructions
我们把截图、旧系统页面、表格、流程图、客服记录、邮件模板和报告样例纳入 Prompt 训练。真实企业里，最有价值的上下文往往不是一句需求，而是已有产物。学员要学会把这些产物转成模型可读、可引用、可验证的上下文。We include screenshots, old system pages, tables, flow maps, support records, email templates, and report samples in prompt training. In real enterprises, the highest-value context is often not a sentence of requirements but existing artifacts. Learners must turn those artifacts into model-readable, referable, and verifiable context.
4. 课程落地：每个项目先做一页工作坊产物4. Course application: every project begins with a workshop artifact
在项目题库里，每个项目都可以要求提交一页工作坊产物：用户是谁、当前流程是什么、关键截图是什么、NSM 是什么、不能做什么、最小 Demo 证明什么。这样学生不会直接跳进代码，而是先建立业务和体验的共同语言。Each project brief can require a one-page workshop artifact: who the users are, what the current workflow is, what the key screenshots are, what the NSM is, what must not be done, and what the minimum demo proves. Learners do not jump directly into code; they first establish a shared business and experience language.
本课训练目标Training Goal
我们把“客户材料整理”作为核心训练项：会问、会看、会截图、会把现场信息变成模型可用上下文。We should treat customer-material preparation as a core skill: ask, observe, capture, and transform field information into model-usable context.
课堂任务Classroom Assignments
为一个企业 AI 项目做一页工作坊画布：用户、流程、截图、NSM、边界、验收。Create a one-page workshop canvas for an enterprise AI project: users, workflow, screenshots, NSM, boundaries, acceptance.
用一张旧系统截图写一个 Prompt，让模型生成页面结构和字段说明。Use one old-system screenshot to write a prompt that asks the model for page structure and field definitions.
把一次访谈记录整理成 Agent 可用的上下文包。Turn one interview transcript into a context pack usable by an agent.
引用Citation
note.com / kamechi_ai_ux. AI / UX practice note on SIer workshops and FDE-style delivery. https://note.com/kamechi_ai_ux/n/n03e655f6d383?hl=en. Accessed 2026-06-28.