本书面向企业管理者、数字化转型负责人和专业技术人员，系统地阐述了DeepSeek 在企业场景中的落地方法论和技术实现指南。全书从部署方法到行业集成再到落地应用，全面覆盖环境搭建与配置、企业应用基础、提示语设计与优化、模型微调与个性化定制、多模态数据处理与融合、分布式训练、与其他企业级专用AI 的结合、与ERP 等企业业务管理系统的结合、创建企业应用AI Agent、创建和集成工作流，以及DeepSeek 应用驱动产业转型升级、面临的挑战与应对等相关议题，旨在帮助读者实现从“能用”迈向“用好”，真正构建可持续的企业AI 能力体系。
高承实，密码学博士，安徽栈谷科技有限公司董事长，深耕于Web 3、人工智能、元宇宙、数字化转型等领域。著有《元宇宙进化逻辑》《区块链技术本质与应用》《回归常识——高博士区块链观察》等系列专著；主编《Web 3中的零知识证明》《区块链中的密码技术》《元宇宙医学》等行业作品，并撰有《与每个老百姓密切相关的稳定币》等科普读物，构建了从学术到普及的完整知识输出体系。同时，发表学术论文20余篇，获得省部级科技进步奖4项。
第1 章 企业应用入门···············································································1
1.1 AI 的发展和DeepSeek 的横空出世····················································2
1.2 DeepSeek 的核心功能和独特优势······················································6
1.3 DeepSeek 在企业中的应用价值······················································.10
1.4 企业部署DeepSeek 需要考虑的因素···············································.12
1.5 企业部署DeepSeek 的主要环节·····················································.26
第2 章 环境搭建与配置·········································································.28
2.1 共性工作··················································································.28
2.2 不同部署方式下的环境搭建与配置·················································.33
2.3 本地部署方式下的环境准备与配置·················································.34
2.4 云端部署方式下的环境准备与配置·················································.40
2.5 企业级部署方式下的环境准备与配置··············································.49
第3 章 企业应用基础············································································.53
3.1 数据准备与预处理······································································.53
3.2 模型训练与评估·········································································.60
3.3 模型部署与应用·········································································.71
第4 章 提示语设计与优化······································································.82
4.1 企业应用环境下提示语设计与优化的作用········································.82
4.2 不同环节下的提示语设计与优化····················································.85
第5 章 模型微调与个性化定制································································.96
5.1 模型微调··················································································.96
5.2 个性化定制方法·········································································106
第6 章 多模态数据处理与融合······························································.112
6.1 数据采集与预处理·····································································.112
6.2 特征提取··················································································121
6.3 模态融合··················································································124
6.4 模型训练与优化·········································································131
第7 章 分布式训练·············································································.137
7.1 计算集群架构············································································137
7.2 并行策略··················································································139
7.3 关键技术优化············································································148
7.4 启动分布式训练·········································································161
第8 章 与企业级专用AI 的结合·····························································.165
8.1 与专用AI 结合的原因·································································165
8.2 与办公协作AI 工具钉钉的结合·····················································169
8.3 与知识管理工具HelpLook 的结合··················································175
8.4 智能客服与对话工具助小咖的结合·················································181
第9 章 与ERP 等企业业务管理系统的结合··············································.189
9.1 与企业业务管理系统结合的特殊性·················································189
9.2 制造业ERP 系统的特点和不足······················································190
9.3 DeepSeek 可以解决的制造业ERP 系统问题······································192
9.4 DeepSeek 与ERP 系统结合的方式、方法和步骤································196
9.5 案例························································································207
第10 章 创建企业应用AI Agent·····························································.209
10.1 什么是AI Agent ·······································································209
10.2 AI Agent 作用发挥的主要领域和适用场景·····································.211
10.3 AI Agent 的结构组成和创建方法··················································214
10.4 基于DeepSeek 创建企业应用AI Agent···········································224
10.5 多Agent 系统··········································································226
10.6 MCP、A2A 和ANP···································································233
第11 章 创建和集成工作流···································································.240
11.1 什么是工作流··········································································240
11.2 工作流的分类··········································································244
11.3 工作流的适用领域和场景···························································247
11.4 创建工作流的一般方法······························································249
11.5 基于DeepSeek 创建企业应用工作流··············································255
11.6 基于DeepSeek 集成工作流··························································263
第12 章 DeepSeek 应用驱动产业转型升级··············································.265
12.1 智能客服系统··········································································265
12.2 质量检测与预测·······································································270
12.3 内容推荐与个性化服务······························································274
12.4 疾病诊断与治疗方案推荐···························································279
12.5 供应链优化与智能调度······························································282
12.6 智能教学辅助系统····································································286
12.7 DeepSeek 及应用驱动产业转型升级··············································290
第13 章 企业应用DeepSeek 面临的挑战与应对·······································.303
13.1 数据安全与隐私保护策略···························································303
13.2 模型幻觉与准确性问题······························································307
13.3 可解释性与逻辑推理能力不足·····················································312
13.4 长文本处理和上下文理解的局限··················································314
13.5 领域适应性与多语言支持问题·····················································316
书单推荐
