AI 辅助软件产品:精选资料(可直接用)¶
本页从全书参考文献与工程常用标准中,提取“最值得优先读/最能直接落地”的资料,并按章节主题归类。目标是让你在需要补背景、补方法或补门禁时,能快速找到权威出处与可执行做法。
P0:先读这些(覆盖全书,优先级最高)¶
- 产品与交付(决策与节奏)
- The Lean Startup(Eric Ries)https://theleanstartup.com/ (对应第 1–2 章:证据与实验门槛;笔记:笔记)
- Continuous Discovery Habits(Teresa Torres)https://www.producttalk.org/continuous-discovery-habits/ (对应第 2/5/19 章:持续发现与实验节奏;笔记:笔记)
- Accelerate(Forsgren/Humble/Kim)https://itrevolution.com/product/accelerate/ (对应第 7/17/18 章:门禁、回归与交付表现;笔记:笔记)
- Continuous Delivery(Humble/Farley)https://martinfowler.com/books/continuousDelivery.html (对应第 7/17 章:发布门禁与回滚;笔记:笔记)
- AI 体验与信任(把不确定性写成可验收)
- Guidelines for Human-AI Interaction(Microsoft)https://www.microsoft.com/en-us/research/publication/guidelines-for-human-ai-interaction/ (对应第 4–6/11 章:提示、纠错、回退与解释;笔记:笔记)
- People + AI Guidebook(Google PAIR)https://pair.withgoogle.com/guidebook/ (对应第 2/4–6/18/20 章:人机协作、反馈回路与评测闭环;笔记:笔记)
- 安全底线(上线必备)
- OWASP ASVS https://owasp.org/www-project-application-security-verification-standard/ (对应第 11/20 章:可验收的安全检查项;笔记:笔记)
- 可靠性门槛(别用‘努力’当承诺)
- SRE:Service Level Objectives(SLO)https://sre.google/sre-book/service-level-objectives/ (对应第 9/17/18 章:SLO 与错误预算;笔记:笔记)
- API 标准(契约与协作)
- OpenAPI Specification https://spec.openapis.org/oas/latest.html (对应第 9/10 章:契约、工具调用与协作;笔记:笔记)
- Spectral(OpenAPI linter)https://github.com/stoplightio/spectral (对应第 3/7/9/10 章:把契约变成 CI 门禁;笔记:笔记)
- 身份与令牌(产品边界的底座)
- OAuth 2.0(RFC 6749)https://www.rfc-editor.org/rfc/rfc6749 (对应第 11 章:认证/授权边界;笔记:笔记)
- OpenID Connect Core 1.0 https://openid.net/specs/openid-connect-core-1_0.html (对应第 11 章:登录互操作;笔记:笔记)
- JWT(RFC 7519)https://www.rfc-editor.org/rfc/rfc7519 (对应第 11 章:令牌契约与审计;笔记:笔记)
- 支付与合规(别等上线才补)
- PCI DSS v4.0 https://www.pcisecuritystandards.org/document_library/?category=pcidss&document=pci_dss (对应第 12/20 章:支付边界与合规;笔记:笔记)
- Usage-based Billing(Stripe)https://docs.stripe.com/billing/subscriptions/usage-based (对应第 12/19 章:计量口径、幂等与对账;笔记:笔记)
第 2 章:需求挖掘与机会判断(Discovery)¶
- The Lean Startup https://theleanstartup.com/ (笔记:笔记)
- Continuous Discovery Habits https://www.producttalk.org/continuous-discovery-habits/ (笔记:笔记)
- People + AI Guidebook https://pair.withgoogle.com/guidebook/ (笔记:笔记)
第 3 章:PRD 与工程合同(Definition)¶
- RFC 2119(MUST/SHOULD/MAY 语义)https://www.rfc-editor.org/rfc/rfc2119 (笔记:笔记)
- OpenAPI Specification https://spec.openapis.org/oas/latest.html (笔记:笔记)
- Spectral(OpenAPI linter)https://github.com/stoplightio/spectral (笔记:笔记)
- ISO/IEC/IEEE 29148(需求工程标准)与 UML(用例建模)属于“写合同”的底层参考:建议后续补充可公开获取的摘要/导读链接,并写成可复用的 PRD 模板(见本仓库
A-templates.md)。
第 4 章:原型与信息架构(Prototyping)¶
- WCAG 2.2 https://www.w3.org/TR/WCAG22/ (笔记:笔记)
- Guidelines for Human-AI Interaction(Microsoft)https://www.microsoft.com/en-us/research/publication/guidelines-for-human-ai-interaction/ (笔记:笔记)
- People + AI Guidebook https://pair.withgoogle.com/guidebook/ (笔记:笔记)
第 6 章:UI 设计:把体验做成可维护资产¶
- Design Tokens 格式 https://www.designtokens.org/TR/drafts/format/ (笔记:笔记)
- axe-core https://github.com/dequelabs/axe-core (笔记:笔记)
- Lighthouse https://developer.chrome.com/docs/lighthouse/ (笔记:笔记)
- Playwright https://playwright.dev/ (笔记:笔记)
- Storybook Test Runner https://storybook.js.org/docs/writing-tests/integrations/test-runner (笔记:笔记)
第 7 章:工程化与编码(Engineering)¶
- Accelerate https://itrevolution.com/product/accelerate/ (笔记:笔记)
- Continuous Delivery https://martinfowler.com/books/continuousDelivery.html (笔记:笔记)
- OpenAPI Specification https://spec.openapis.org/oas/latest.html (笔记:笔记)
- Spectral(OpenAPI linter)https://github.com/stoplightio/spectral (笔记:笔记)
第 9 章:后端架构(契约/错误语义/可靠性)¶
- HTTP Semantics(RFC 9110)https://www.rfc-editor.org/rfc/rfc9110 (笔记:笔记)
- Problem Details for HTTP APIs(RFC 9457)https://www.rfc-editor.org/rfc/rfc9457 (笔记:笔记)
- SRE:Service Level Objectives(SLO)https://sre.google/sre-book/service-level-objectives/ (笔记:笔记)
第 10 章:Agent 架构与 RAG(智能层)¶
- RAG 原始论文(Lewis et al., 2020)https://arxiv.org/abs/2005.11401 (笔记:笔记)
- BM25 综述(Robertson & Zaragoza, 2009)https://www.nowpublishers.com/article/Details/INR-019 (笔记:笔记)
- FAISS(Johnson et al., 2019)https://arxiv.org/abs/1702.08734 (笔记:笔记)
- RAGAS(Shahul et al., 2023)https://arxiv.org/abs/2309.15217 (笔记:笔记)
- ReAct(Yao et al., 2023)https://arxiv.org/abs/2210.03629 (笔记:笔记)
- LangGraph 文档 https://docs.langchain.com/oss/python/langgraph/overview/ (笔记:笔记)
- AutoGen(Wu et al., 2023)https://arxiv.org/abs/2308.08155 (笔记:笔记)
- OpenAI Tool Use / Function Calling(文档入口)https://platform.openai.com/docs/ (笔记:笔记)
第 11 章:用户模块(AuthN/AuthZ/审计)¶
- OAuth 2.0(RFC 6749)https://www.rfc-editor.org/rfc/rfc6749 (笔记:笔记)
- OpenID Connect Core 1.0 https://openid.net/specs/openid-connect-core-1_0.html (笔记:笔记)
- JWT(RFC 7519)https://www.rfc-editor.org/rfc/rfc7519 (笔记:笔记)
- OWASP ASVS https://owasp.org/www-project-application-security-verification-standard/ (笔记:笔记)
第 12 章:付费模块(计量/账本/对账/风控)¶
- PCI DSS v4.0 https://www.pcisecuritystandards.org/document_library/?category=pcidss&document=pci_dss (笔记:笔记)
- Usage-based Billing(Stripe)https://docs.stripe.com/billing/subscriptions/usage-based (笔记:笔记)
第 13 章:数据收集与清洗(Data)¶
- Datasheets for Datasets(Gebru et al., 2021)https://arxiv.org/abs/1803.09010 (笔记:笔记)
- LLM 数据去重(Lee et al., 2021)https://arxiv.org/abs/2107.06499 (笔记:笔记)
第 15 章:后训练与对齐(Post-train / Alignment)¶
- LoRA(Hu et al., 2021/2022)https://arxiv.org/abs/2106.09685 (笔记:笔记)
- Self-Instruct(Wang et al., 2023)https://arxiv.org/abs/2212.10560 (笔记:笔记)
- WizardLM / Evol-Instruct(Xu et al., 2023)https://arxiv.org/abs/2304.12244 (笔记:笔记)
- QLoRA(Dettmers et al., 2023)https://arxiv.org/abs/2305.14314 (笔记:笔记)
- RLHF(Ouyang et al., 2022)https://arxiv.org/abs/2203.02155 (笔记:笔记)
- DPO(Rafailov et al., 2023)https://arxiv.org/abs/2305.18290 (笔记:笔记)
- PPO(Schulman et al., 2017)https://arxiv.org/abs/1707.06347 (笔记:笔记)
- Deep RL from Human Preferences(Christiano et al., 2017)https://arxiv.org/abs/1706.03741 (笔记:笔记)
- Learning to Summarize from Human Feedback(Stiennon et al., 2020)https://arxiv.org/abs/2009.01325 (笔记:笔记)
- Constitutional AI(Bai et al., 2022)https://arxiv.org/abs/2212.08073 (笔记:笔记)
第 16 章:推理优化(Inference)¶
- vLLM(Kwon et al., 2023)https://arxiv.org/abs/2309.06180 (笔记:笔记)
- TensorRT-LLM(官方资料入口)https://github.com/NVIDIA/TensorRT-LLM (笔记:笔记)
- GPTQ(Frantar et al., 2024)https://arxiv.org/abs/2210.17323 (笔记:笔记)
- AWQ(Lin et al., 2023)https://arxiv.org/abs/2306.00978 (笔记:笔记)
- Text Generation Inference(TGI)https://github.com/huggingface/text-generation-inference (笔记:笔记)
- KServe(推理服务编排)https://kserve.github.io/website/ (笔记:笔记)
第 17–18 章:部署/运维与评测(Ops & Evaluation)¶
- OpenTelemetry https://opentelemetry.io/docs/ (笔记:笔记)
- LLM-as-a-Judge(Zheng et al., 2023)https://arxiv.org/abs/2306.05685 (笔记:笔记)
- Prometheus https://prometheus.io/docs/introduction/overview/ (笔记:笔记)
- Grafana https://grafana.com/docs/grafana/latest/ (笔记:笔记)
- SRE:Service Level Objectives(SLO)https://sre.google/sre-book/service-level-objectives/ (笔记:笔记)
- Guardrails(Building Guardrails for LLMs, 2024)https://arxiv.org/abs/2402.01822 (笔记:笔记)
第 19 章:迭代与增长(路线图/实验/定价)¶
- Continuous Discovery Habits https://www.producttalk.org/continuous-discovery-habits/ (笔记:笔记)
- Usage-based Billing(Stripe)https://docs.stripe.com/billing/subscriptions/usage-based (笔记:笔记)
第 20 章:合规与伦理(风险/隐私/安全/治理)¶
- NIST AI RMF https://www.nist.gov/itl/ai-risk-management-framework (笔记:笔记)
- Concrete Problems in AI Safety(Amodei et al., 2016)https://arxiv.org/abs/1606.06565 (笔记:笔记)
说明:如何“查找与补充”更多材料¶
- 想按章节找:先看
../indexes.md的“按章节适配”分组,再回到对应notes/和deepresearch/。 - 想补某一章:优先补“权威标准/协议/论文的入口页”,其次补“可执行的工程手册”,最后再补“观点型文章”。
- 如果你发现某一章资料少:先补“底层标准”(例如 Auth 的 RFC、支付的合规标准),再补“可操作实践”(例如 runbook、回归策略、对账流程)。