本書面向企業(yè)管理者、數(shù)字化轉型負責人和專業(yè)技術人員，系統(tǒng)地闡述了DeepSeek 在企業(yè)場景中的落地方法論和技術實現(xiàn)指南。全書從部署方法到行業(yè)集成再到落地應用，全面覆蓋環(huán)境搭建與配置、企業(yè)應用基礎、提示語設計與優(yōu)化、模型微調與個性化定制、多模態(tài)數(shù)據(jù)處理與融合、分布式訓練、與其他企業(yè)級專用AI 的結合、與ERP 等企業(yè)業(yè)務管理系統(tǒng)的結合、創(chuàng)建企業(yè)應用AI Agent、創(chuàng)建和集成工作流，以及DeepSeek 應用驅動產(chǎn)業(yè)轉型升級、面臨的挑戰(zhàn)與應對等相關議題，旨在幫助讀者實現(xiàn)從“能用”邁向“用好”，真正構建可持續(xù)的企業(yè)AI 能力體系。
高承實，密碼學博士，安徽棧谷科技有限公司董事長，深耕于Web 3、人工智能、元宇宙、數(shù)字化轉型等領域。著有《元宇宙進化邏輯》《區(qū)塊鏈技術本質與應用》《回歸常識——高博士區(qū)塊鏈觀察》等系列專著；主編《Web 3中的零知識證明》《區(qū)塊鏈中的密碼技術》《元宇宙醫(yī)學》等行業(yè)作品，并撰有《與每個老百姓密切相關的穩(wěn)定幣》等科普讀物，構建了從學術到普及的完整知識輸出體系。同時，發(fā)表學術論文20余篇，獲得省部級科技進步獎4項。
第1 章 企業(yè)應用入門···············································································1
1.1 AI 的發(fā)展和DeepSeek 的橫空出世····················································2
1.2 DeepSeek 的核心功能和獨特優(yōu)勢······················································6
1.3 DeepSeek 在企業(yè)中的應用價值······················································.10
1.4 企業(yè)部署DeepSeek 需要考慮的因素···············································.12
1.5 企業(yè)部署DeepSeek 的主要環(huán)節(jié)·····················································.26
第2 章 環(huán)境搭建與配置·········································································.28
2.1 共性工作··················································································.28
2.2 不同部署方式下的環(huán)境搭建與配置·················································.33
2.3 本地部署方式下的環(huán)境準備與配置·················································.34
2.4 云端部署方式下的環(huán)境準備與配置·················································.40
2.5 企業(yè)級部署方式下的環(huán)境準備與配置··············································.49
第3 章 企業(yè)應用基礎············································································.53
3.1 數(shù)據(jù)準備與預處理······································································.53
3.2 模型訓練與評估·········································································.60
3.3 模型部署與應用·········································································.71
第4 章 提示語設計與優(yōu)化······································································.82
4.1 企業(yè)應用環(huán)境下提示語設計與優(yōu)化的作用········································.82
4.2 不同環(huán)節(jié)下的提示語設計與優(yōu)化····················································.85
第5 章 模型微調與個性化定制································································.96
5.1 模型微調··················································································.96
5.2 個性化定制方法·········································································106
第6 章 多模態(tài)數(shù)據(jù)處理與融合······························································.112
6.1 數(shù)據(jù)采集與預處理·····································································.112
6.2 特征提取··················································································121
6.3 模態(tài)融合··················································································124
6.4 模型訓練與優(yōu)化·········································································131
第7 章 分布式訓練·············································································.137
7.1 計算集群架構············································································137
7.2 并行策略··················································································139
7.3 關鍵技術優(yōu)化············································································148
7.4 啟動分布式訓練·········································································161
第8 章 與企業(yè)級專用AI 的結合·····························································.165
8.1 與專用AI 結合的原因·································································165
8.2 與辦公協(xié)作AI 工具釘釘?shù)慕Y合·····················································169
8.3 與知識管理工具HelpLook 的結合··················································175
8.4 智能客服與對話工具助小咖的結合·················································181
第9 章 與ERP 等企業(yè)業(yè)務管理系統(tǒng)的結合··············································.189
9.1 與企業(yè)業(yè)務管理系統(tǒng)結合的特殊性·················································189
9.2 制造業(yè)ERP 系統(tǒng)的特點和不足······················································190
9.3 DeepSeek 可以解決的制造業(yè)ERP 系統(tǒng)問題······································192
9.4 DeepSeek 與ERP 系統(tǒng)結合的方式、方法和步驟································196
9.5 案例························································································207
第10 章 創(chuàng)建企業(yè)應用AI Agent·····························································.209
10.1 什么是AI Agent ·······································································209
10.2 AI Agent 作用發(fā)揮的主要領域和適用場景·····································.211
10.3 AI Agent 的結構組成和創(chuàng)建方法··················································214
10.4 基于DeepSeek 創(chuàng)建企業(yè)應用AI Agent···········································224
10.5 多Agent 系統(tǒng)··········································································226
10.6 MCP、A2A 和ANP···································································233
第11 章 創(chuàng)建和集成工作流···································································.240
11.1 什么是工作流··········································································240
11.2 工作流的分類··········································································244
11.3 工作流的適用領域和場景···························································247
11.4 創(chuàng)建工作流的一般方法······························································249
11.5 基于DeepSeek 創(chuàng)建企業(yè)應用工作流··············································255
11.6 基于DeepSeek 集成工作流··························································263
第12 章 DeepSeek 應用驅動產(chǎn)業(yè)轉型升級··············································.265
12.1 智能客服系統(tǒng)··········································································265
12.2 質量檢測與預測·······································································270
12.3 內容推薦與個性化服務······························································274
12.4 疾病診斷與治療方案推薦···························································279
12.5 供應鏈優(yōu)化與智能調度······························································282
12.6 智能教學輔助系統(tǒng)····································································286
12.7 DeepSeek 及應用驅動產(chǎn)業(yè)轉型升級··············································290
第13 章 企業(yè)應用DeepSeek 面臨的挑戰(zhàn)與應對·······································.303
13.1 數(shù)據(jù)安全與隱私保護策略···························································303
13.2 模型幻覺與準確性問題······························································307
13.3 可解釋性與邏輯推理能力不足·····················································312
13.4 長文本處理和上下文理解的局限··················································314
13.5 領域適應性與多語言支持問題·····················································316
書單推薦
