[{"data":1,"prerenderedAt":1690},["ShallowReactive",2],{"site-content":3,"content-source":1688},{"common":4,"brand":31,"navigation":35,"footerLinks":54,"links":82,"home":102,"serviceCards":155,"publicServiceCards":185,"cases":202,"testimonials":570,"faqs":592,"about":617,"productsPage":675,"team":698,"devHub":734,"investorRelations":780,"demosPage":808,"publicServicesPage":836,"aiHelp":851,"linksPage":860,"productCards":868,"demoConcepts":950,"productDetails":960,"demos":1211,"detailPages":1529},{"skipToContent":5,"footerTitle":6,"deliveryFocus":7,"detailLabels":8,"demoLabels":13,"conversionCta":17,"theme":28},"跳到主要内容","友情链接","交付聚焦",{"overview":9,"features":10,"scenarios":11,"advantages":12},"服务概述","主要特性","应用场景","技术优势",{"stageTitle":14,"openOriginal":15,"missing":16},"阶段说明","在新窗口打开原始演示","未找到该演示资源。",{"eyebrow":18,"stages":19,"primaryLabel":23,"secondaryCasesLabel":24,"secondaryProductsLabel":25,"secondaryDemosLabel":26,"tertiaryLabel":27},"Next Step",[20,21,22],"需求沟通与场景梳理","方案确认与路径选择","进入开发或交付阶段","发起咨询","查看案例","查看产品","查看演示","邮件直连",{"light":29,"dark":30},"亮色","暗色",{"name":32,"tagline":33,"logo":34},"开发喵 AI","AI 服务与区块链解决方案","https:\u002F\u002Fcdn.sa.net\u002F2024\u002F07\u002F17\u002FyPZoEwn6uWFxihJ.jpg",[36,39,42,45,48,51],{"label":37,"to":38},"首页","\u002F",{"label":40,"to":41},"产品","\u002Fproducts",{"label":43,"to":44},"演示","\u002Faidemo",{"label":46,"to":47},"团队","\u002Fteam",{"label":49,"to":50},"开发者","\u002Fdev-hub",{"label":52,"to":53},"关于","\u002Fabout",[55,59,62,65,68,71,74,77,80],{"label":56,"to":57,"external":58},"OpenLens","https:\u002F\u002Fopenlens.pro",true,{"label":60,"to":61,"external":58},"HandyHub","https:\u002F\u002Fhandyhub.vip",{"label":63,"to":64,"external":58},"KFMPing","https:\u002F\u002Fkfmping.cc",{"label":66,"to":67,"external":58},"云创动力","https:\u002F\u002Fycdledu.com",{"label":69,"to":70,"external":58},"六行.神算","https:\u002F\u002Fwww.openai-ss.cn\u002F",{"label":72,"to":73,"external":58},"AI喵帮助","https:\u002F\u002Fhelp.1155665ai.com",{"label":75,"to":76,"external":58},"用户中心","https:\u002F\u002Fkfmhome.com",{"label":78,"to":79},"公共服务平台","\u002Fpublic-services",{"label":81,"to":50},"开发者中心",[83,87,92,97],{"title":84,"description":85,"icon":86,"to":38},"返回主页","回到站点总入口，查看服务能力与最新内容。","Home",{"title":88,"description":89,"icon":90,"to":91,"external":58},"GitHub","查看团队主页与开源协作入口。","Github","https:\u002F\u002Fgithub.com\u002Fkaifamiao",{"title":93,"description":94,"icon":95,"to":96},"AI 服务","API、大模型、知识库、RAG、Agent 等模块入口。","Sparkles","\u002F#services",{"title":98,"description":99,"icon":100,"to":101},"区块链服务","智能合约与 DApp 等说明见专题页。","Blocks","\u002Fblockchain",{"headTitle":33,"hero":103,"services":122,"publicServices":137,"cases":142,"faq":147,"feedback":152},{"eyebrow":104,"title":105,"description":106,"primary":107,"secondary":108,"stackLabel":110,"stackItems":111,"stats":115},"AI That Everyone Can Use","让 AI 融入业务、产品与日常工作。","开发喵 AI 聚焦 API、多模型、知识库、RAG、Agent 与区块链解决方案，帮助团队更快完成从验证到上线。",{"label":25,"to":41},{"label":109,"to":44},"体验原理演示","Capability Stack",[112,113,114],"多模型 API 与路由策略","知识库、RAG 与 Agent 工程","公共服务平台与配套文档",[116,119],{"value":117,"label":118},"6+","核心服务模块",{"value":120,"label":121},"4","公共服务能力",{"title":123,"description":124,"metrics":125},"服务能力","按模块浏览，进入详情页查看交付边界与应用场景。",[126,129,132,135],{"value":127,"label":128},"100+","服务企业",{"value":130,"label":131},"10000+","个人用户",{"value":133,"label":134},"5","覆盖行业",{"value":117,"label":136},"核心服务",{"eyebrow":138,"title":78,"description":139,"cta":140},"Public Services","统一整理开通流程、Key 使用方式与服务明细，适合研发团队集中接入。",{"label":141,"to":79},"查看开通流程",{"title":143,"description":144,"resultLabel":145,"listLabel":146},"客户案例","横向滑动查看；数字为项目摘要，非营销口号堆砌。","成果摘要","查看全部案例",{"title":148,"description":149,"moreLabel":150,"pageTitle":151},"常见问题","关于接入、交付、周期与合作方式的高频问题。","更多问题","全部常见问题",{"title":153,"description":154},"客户反馈","用行业、项目类型和交付结果去说明合作价值，而不是只放几句匿名评价。",[156,162,167,172,177,182],{"title":157,"description":158,"icon":159,"to":160,"cta":161},"AI API 服务","稳定的多模型调用与接入方案，便于现有系统快速获得推理与生成能力。","Zap","\u002Fai-api","了解详情",{"title":163,"description":164,"icon":165,"to":166,"cta":161},"大模型开发与训练","架构设计、数据与训练管线、微调与评估，覆盖从实验到上线的工程环节。","Cpu","\u002Fllm-development",{"title":168,"description":169,"icon":170,"to":171,"cta":161},"AI 知识库搭建","文档治理、索引与权限，把分散资料变成可检索、可审计的知识资产。","BookOpen","\u002Fai-knowledge-base",{"title":173,"description":174,"icon":175,"to":176,"cta":161},"RAG 应用开发","检索增强生成，更可控的问答与内容生成，贴合业务语料与合规要求。","FileSearch","\u002Frag-app",{"title":178,"description":179,"icon":180,"to":181,"cta":161},"Agent 应用开发","为复杂流程设计可协作、可执行、可观测的自主式 AI Agent 系统。","Bot","\u002Fagent-app",{"title":183,"description":184,"icon":100,"to":101,"cta":161},"区块链开发","智能合约、链上交互、钱包接入与 DApp 方案，覆盖链上产品交付。",[186,190,194,198],{"title":187,"description":188,"icon":189},"开发喵短信服务","用于验证码、通知与营销消息触达，支持回执和状态查询。","MessageSquareHeart",{"title":191,"description":192,"icon":193},"大模型接入服务","统一路由多模型能力，支持配额管理、日志与调用观测。","BrainCircuit",{"title":195,"description":196,"icon":197},"科学网络转化服务","提升跨区域网络连接质量，满足研发、抓取和数据传输场景。","Network",{"title":199,"description":200,"icon":201},"IP 质量检测服务","检测 IP 可用性、风险状态与质量评分，辅助前置筛选。","ShieldCheck",[203,243,280,317,354,391,428,463,499,535],{"id":204,"index":204,"slug":205,"title":206,"company":207,"description":208,"outcomes":209,"category":214,"industry":215,"duration":216,"metrics":217,"summary":226,"challenge":227,"solution":231,"architecture":235,"results":239},1,"enterprise-customer-support","企业智能客服系统","某大型科技公司","基于大模型的客服系统：分流、话术与质检一体化，缩短响应链路。",[210,211,212,213],"响应时间显著缩短","满意度与解决率提升","人工坐席负载下降","7×24 自动覆盖","enterprise","企业服务 \u002F 客服运营","8 周",[218,220,223],{"label":219,"value":216},"交付周期",{"label":221,"value":222},"服务覆盖","7×24",{"label":224,"value":225},"核心结果","响应链路缩短","围绕海量咨询入口、复杂业务口径和服务质检压力，建设一套结合大模型问答、知识检索和人工兜底的客服系统。",[228,229,230],"客服入口分散，用户问题类型多，人工分流效率低。","业务知识更新频繁，旧知识库维护成本高，答案一致性差。","客服质检依赖人工抽检，覆盖率低，无法形成稳定优化闭环。",[232,233,234],"搭建统一的意图识别与分流入口，把高频问题优先交给 AI 处理。","把产品文档、流程说明和常见问题纳入知识库，并通过 RAG 提升回答可控性。","增加质检规则与会话摘要能力，让运营团队快速定位异常对话。",[236,237,238],"多渠道接入层","知识库与检索增强问答","人工转接与质检分析模块",[240,241,242],"高频重复咨询由 AI 稳定承接，人工坐席只处理复杂个案。","客服响应链路明显缩短，业务口径更统一。","质检从抽样走向高覆盖，后续培训和优化更有依据。",{"id":244,"index":244,"slug":245,"title":246,"company":247,"description":248,"outcomes":249,"category":254,"industry":255,"duration":256,"metrics":257,"summary":263,"challenge":264,"solution":268,"architecture":272,"results":276},2,"intelligent-knowledge-platform","智能知识库平台","知名教育机构","教学资源入库、版本与权限管理，支持教研场景下的精准检索。",[250,251,252,253],"检索效率提升","多格式文档统一治理","问答与引用可追溯","使用率持续增长","knowledge","教育 \u002F 教研管理","6 周",[258,259,262],{"label":219,"value":256},{"label":260,"value":261},"内容类型","多格式文档",{"label":224,"value":250},"为教研团队构建统一知识底座，解决教学资料分散、版本混乱和跨角色检索效率低的问题。",[265,266,267],"历史资料分散在网盘、群文件和本地目录，统一治理难度高。","教师、教研、管理人员对资料的权限要求不同。","传统关键词搜索难以满足复杂问题查询。",[269,270,271],"对教案、课件、制度和培训材料建立统一入库规范。","引入文档切片、向量检索和权限过滤机制，保证可查也可控。","问答结果附带引用片段与来源，提升教研使用信任度。",[273,274,275],"文档治理与解析","向量索引与权限控制","问答与引用展示层",[277,278,279],"资料检索入口统一，教研团队不再反复问人找文件。","权限边界清晰，敏感内容不会误暴露给无关角色。","问答引用可回溯，知识使用更规范。",{"id":281,"index":281,"slug":282,"title":283,"company":284,"description":285,"outcomes":286,"category":291,"industry":292,"duration":293,"metrics":294,"summary":300,"challenge":301,"solution":305,"architecture":309,"results":313},3,"ai-agent-automation","AI Agent 自动化","互联网企业","运营与内部流程的 Agent 编排，关键步骤保留人工审核节点。",[287,288,289,290],"重复劳动自动化","人为差错减少","复杂分支可配置","ROI 可量化复盘","automation","互联网 \u002F 运营自动化","10 周",[295,296,299],{"label":219,"value":293},{"label":297,"value":298},"流程形态","多步骤编排",{"label":224,"value":287},"为运营和内部协作流程建设 Agent 执行闭环，让系统能拆任务、调工具、回读结果并在关键节点交给人工确认。",[302,303,304],"内部流程跨系统、跨角色，人工执行步骤多且容易漏。","一次性脚本难以应对复杂分支和异常状态。","完全自动化风险高，需要保留人工审核能力。",[306,307,308],"将任务拆成若干明确节点，由 Agent 逐步执行并记录状态。","为不同任务接入查询、通知、表单和内部数据工具。","在人审节点引入回退和继续机制，避免误操作。",[310,311,312],"任务编排引擎","工具调用层","状态持久化与人审流程",[314,315,316],"重复性运营流程大幅减少人工介入。","异常路径可观测、可追溯，便于复盘与优化。","业务团队更愿意接受 Agent 协助，而不是担心黑盒执行。",{"id":318,"index":318,"slug":319,"title":320,"company":321,"description":322,"outcomes":323,"category":254,"industry":328,"duration":329,"metrics":330,"summary":337,"challenge":338,"solution":342,"architecture":346,"results":350},4,"enterprise-rag-qa","企业 RAG 问答","金融机构","内规、研报与制度库接入 RAG，回答带出处，满足合规查阅习惯。",[324,325,326,327],"答案准确率提升","延迟与吞吐优化","多语言资料支持","知识覆盖扩展","金融 \u002F 合规与研究","7 周",[331,332,335],{"label":219,"value":329},{"label":333,"value":334},"关键能力","带出处回答",{"label":224,"value":336},"准确率提升","把分散的制度、内规和研究资料连接到可追溯问答系统，重点解决回答可信度和出处要求。",[339,340,341],"用户需要的不只是答案，还要能看到来源与依据。","文档体量大且更新快，纯人工维护索引困难。","对准确性和可控性的要求高于一般内容问答场景。",[343,344,345],"设计适配金融资料的切片和召回策略，减少无关片段噪音。","在回答中显示引用来源、标题与片段，满足合规查阅习惯。","配合重排和上下文治理，稳定高价值资料的命中率。",[347,348,349],"文档解析与向量检索","重排与上下文拼接","带出处的回答生成",[351,352,353],"业务用户能更快找到制度依据和研究结论。","答案可解释性明显增强，内部使用阻力降低。","多类资料被统一纳入可维护的知识问答体系。",{"id":355,"index":355,"slug":356,"title":357,"company":358,"description":359,"outcomes":360,"category":365,"industry":366,"duration":367,"metrics":368,"summary":374,"challenge":375,"solution":379,"architecture":383,"results":387},5,"smart-contract-platform","智能合约平台","区块链公司","合约模板、审计协作与多环境部署，面向业务方的可读抽象层。",[361,362,363,364],"执行路径优化","Gas 与成本评估","安全基线对齐","多链策略支持","platform","区块链 \u002F 链上产品","9 周",[369,370,373],{"label":219,"value":367},{"label":371,"value":372},"部署范围","多环境 \u002F 多链",{"label":224,"value":363},"围绕合约模板管理、部署流程和审计协作，降低业务团队接触底层链上实现的复杂度。",[376,377,378],"业务方理解合约逻辑成本高，需求沟通效率低。","多环境部署和多链兼容带来版本管理压力。","安全和审计要求使上线流程更复杂。",[380,381,382],"抽象常用合约能力，形成模板化交付方式。","建立测试、部署和审计协作流程，减少线上风险。","为业务方提供更可读的配置和状态展示。",[384,385,386],"合约模板层","测试与部署流水线","审计协作与监控面板",[388,389,390],"业务方对链上逻辑的理解成本明显下降。","多环境与多链部署流程更规范。","上线前风险暴露更早，协作效率更高。",{"id":392,"index":392,"slug":393,"title":394,"company":395,"description":396,"outcomes":397,"category":402,"industry":403,"duration":404,"metrics":405,"summary":411,"challenge":412,"solution":416,"architecture":420,"results":424},6,"custom-model-training","大模型定制训练","AI 创新公司","领域数据清洗、训练预算与评测指标对齐，避免“训完不可用”。",[398,399,400,401],"任务指标提升","训练周期压缩","算力成本可控","持续迭代机制","training","AI \u002F 行业模型","12 周",[406,407,410],{"label":219,"value":404},{"label":408,"value":409},"工作重点","训练与评测闭环",{"label":224,"value":398},"围绕领域任务构建训练数据、评估标准和推理上线方案，让训练结果真正服务业务指标。",[413,414,415],"业务目标和训练指标之间存在脱节。","领域数据质量参差不齐，直接训练效果不稳定。","算力预算有限，需要在性能和成本间做平衡。",[417,418,419],"从业务任务反推训练目标，明确数据与评测口径。","清洗指令数据并建立分阶段评估机制。","结合推理优化方案，避免训练成果难以上线。",[421,422,423],"数据治理与标注","训练与评测流水线","推理优化与上线方案",[425,426,427],"模型在目标任务上的有效指标得到提升。","训练和评测形成可复用流程，后续迭代更稳。","算力投入和产出关系更清晰。",{"id":429,"index":429,"slug":430,"title":431,"company":432,"description":433,"outcomes":434,"category":365,"industry":439,"duration":216,"metrics":440,"summary":446,"challenge":447,"solution":451,"architecture":455,"results":459},7,"ai-api-platform","AI API 服务平台","SaaS 服务商","统一网关、配额与计费观测，多模型路由对下游产品透明。",[435,436,437,438],"调用量增长","P99 延迟下降","多模型统一接入","高可用架构","SaaS \u002F 平台能力",[441,442,445],{"label":219,"value":216},{"label":443,"value":444},"核心能力","统一模型网关",{"label":224,"value":436},"为多产品线统一建设 AI API 平台，解决供应商切换、配额管理和计费观测问题。",[448,449,450],"多个产品分别对接模型，维护成本高。","不同模型供应商接口差异大，切换困难。","调用观测和成本控制缺乏统一平台。",[452,453,454],"建立统一 API 网关和模型路由层。","对接日志、配额、计费与监控，向下游提供稳定接口。","通过策略路由提升高峰期可用性和响应表现。",[456,457,458],"统一鉴权网关","模型路由策略层","日志计费与监控看板",[460,461,462],"新业务接入速度明显提升。","模型切换与扩展不再影响下游产品接口。","成本、延迟和调用量都能统一观测。",{"id":464,"index":464,"slug":465,"title":466,"company":467,"description":468,"outcomes":469,"category":254,"industry":474,"duration":293,"metrics":475,"summary":482,"challenge":483,"solution":487,"architecture":491,"results":495},8,"university-knowledge-base-system","某高校知识库系统","某高校","围绕校内制度、科研资料与教学文档建设统一知识库系统，支持权限检索与来源追溯。",[470,471,472,473],"校内资料统一入库","师生检索效率提升","知识权限分层管理","答案引用可追溯","高校 \u002F 信息化建设",[476,477,480],{"label":219,"value":293},{"label":478,"value":479},"服务对象","师生与管理部门",{"label":224,"value":481},"资料统一入库","围绕高校制度文件、科研资料和教学文档，建设面向多角色的统一知识库系统。",[484,485,486],"校内资料类型多，来源分散，历史沉淀缺乏统一入口。","教师、学生和管理部门对资料访问权限差异明显。","传统搜索只能找文件名，难以回答复杂问题。",[488,489,490],"统一接入制度、科研、教学三类核心资料，并做分类治理。","按角色与部门划分权限，检索结果自动过滤无权限内容。","通过知识问答与来源引用，让师生能直接定位到相关原文。",[492,493,494],"多源资料接入与治理","权限检索与向量索引","问答与引用展示",[496,497,498],"校内知识资产不再分散在多个系统和个人目录中。","教师和学生可以更快找到制度依据与教学资料。","问答结果可追溯到来源，信息可信度更高。",{"id":500,"index":500,"slug":501,"title":502,"company":503,"description":504,"outcomes":505,"category":365,"industry":509,"duration":510,"metrics":511,"summary":518,"challenge":519,"solution":523,"architecture":527,"results":531},9,"listed-company-multi-model-platform","某上市公司大模型聚合平台","某上市公司","整合多家模型供应商能力，统一提供调用路由、监控、配额与业务接入界面。",[437,506,507,508],"调用成本更可控","平台监控能力增强","业务团队接入更快","企业平台 \u002F AI 中台","11 周",[512,513,516],{"label":219,"value":510},{"label":514,"value":515},"平台角色","企业 AI 中台",{"label":224,"value":517},"接入速度提升","建设面向内部业务团队的大模型聚合平台，让模型供应商能力统一收敛在一个平台层中。",[520,521,522],"不同业务团队重复对接模型，成本和风险都高。","模型供应商能力和计费方式不统一，平台治理难度大。","缺少统一监控与策略配置，无法快速做供应商切换。",[524,525,526],"抽象统一调用协议，把底层多家模型供应商做成平台能力。","建设统一监控、配额、日志和成本分析能力。","通过路由策略与配置管理提升平台的灵活性和可控性。",[528,529,530],"统一调用网关","模型供应商适配层","监控配额与成本分析",[532,533,534],"业务团队接入模型的门槛和时间显著下降。","平台具备成本观察和模型切换能力。","多模型能力开始从“分散接入”变成“平台资产”。",{"id":536,"index":536,"slug":537,"title":538,"company":539,"description":540,"outcomes":541,"category":291,"industry":546,"duration":367,"metrics":547,"summary":553,"challenge":554,"solution":558,"architecture":562,"results":566},10,"discipline-research-report-system","某纪律组织调研报告系统","某纪律组织","面向调研与报告撰写场景，打通资料沉淀、检索问答、报告生成与版本管理流程。",[542,543,544,545],"调研资料集中治理","报告草拟效率提升","版本流转更规范","过程留痕可审计","政务 \u002F 调研与报告",[548,549,551],{"label":219,"value":367},{"label":11,"value":550},"调研与报告流转",{"label":224,"value":552},"初稿效率提升","围绕调研资料沉淀、报告草拟和版本流转建设一套可留痕、可审计的报告系统。",[555,556,557],"调研资料积累分散，后续复用效率低。","报告撰写依赖人工整理资料，初稿效率受限。","版本流转和修改记录不够规范，追踪成本高。",[559,560,561],"统一管理调研资料、历史报告和参考素材。","引入知识问答与报告生成辅助，提升初稿整理效率。","把版本管理和流转记录纳入系统，满足审计与留痕要求。",[563,564,565],"资料治理与知识检索","报告生成辅助","版本流转与审计记录",[567,568,569],"调研资料和历史成果能被持续复用。","报告初稿生成效率明显提升。","版本流转更加规范，可审计性更强。",[571,578,585],{"industry":572,"projectType":573,"result":574,"quote":575,"author":576,"meta":577},"上市公司","大模型聚合平台","多模型统一接入，业务团队接入效率提升","需求理解很快，能把技术方案翻译成业务团队能直接判断的交付路径。","某上市公司产品负责人","平台建设项目",{"industry":579,"projectType":580,"result":581,"quote":582,"author":583,"meta":584},"高校","知识库系统","资料检索效率提升，引用来源更清晰","知识库系统上线后，老师和学生检索资料的效率明显提升，引用来源也更清晰。","某高校信息化负责人","校内知识治理项目",{"industry":586,"projectType":587,"result":588,"quote":589,"author":590,"meta":591},"纪律组织","调研报告系统","报告草拟效率提升，流程留痕更规范","项目推进节奏稳，文档和沟通都比较规范，后续扩展也预留了空间。","某纪律组织项目负责人","调研与报告项目",[593,596,599,602,605,608,611,614],{"q":594,"a":595},"你们支持哪些交付方式？","支持咨询+开发、联合研发、按模块交付，以及长期运维协作。",{"q":597,"a":598},"从需求到上线一般要多久？","PoC 通常 1-3 周，正式版本视范围通常 4-12 周可上线首版。",{"q":600,"a":601},"可以私有化部署吗？","可以，支持云端与本地私有化部署，并可对接企业 IAM 与审计体系。",{"q":603,"a":604},"如何申请公共服务 Key？","提交需求后完成审核即可发放 Key，详情可查看",{"q":606,"a":607},"你们安全可靠吗？","我们长期深耕 AI 领域，已服务多类行业客户，重视稳定交付与长期合作。公司合规经营，严格按照业务规范与法律要求开展服务，确保合作可持续、可追溯。",{"q":609,"a":610},"你们在海外有业务吗？","有。我们支持海外客户与跨区域项目协作，具备多时区沟通与全球化交付经验，可按场景提供云端、网络与模型接入方案。",{"q":612,"a":613},"你们公司在哪里？","公司位于天府之国（成都），并支持远程协作与跨城市项目交付。",{"q":615,"a":616},"你们主要的客户是谁？","我们主要服务政府、央企\u002F国企及事业单位，同时也支持具备明确业务目标的企业客户，强调安全合规、稳定交付与长期维护。",{"headTitle":618,"hero":619,"missionTitle":626,"mission":627,"whyTitle":630,"whyDescription":631,"whyNeeds":632,"capabilitiesTitle":652,"capabilities":653,"visionTitle":671,"vision":672},"关于我们",{"eyebrow":620,"title":621,"description":622,"primary":623,"secondary":624},"About","关于开发喵 AI","我们希望让 AI 融入生活，让更多人不受限制地使用 AI，把复杂能力做成真正可用的产品与服务。",{"label":25,"to":41},{"label":625,"to":50},"联系团队","使命",[628,629],"让 AI 融入生活，让更多人不受限制地使用 AI。","“AI that everyone can use” 是我们的口号，也是站点和产品设计的核心方向。","为什么需要我们？","很多团队不是不知道 AI 有价值，而是不确定怎么在合规、安全、成本和本地化限制下真正落地。我们的价值，在于把这些限制条件变成可交付方案。",[633,636,640,644,648],{"title":634,"description":635,"icon":201},"合规需求","很多行业在接入 AI 时首先遇到的不是模型效果，而是合规边界。我们会结合业务场景梳理数据流向、权限边界、供应商选择和落地方式，帮助团队在合规框架内推进，而不是停留在“想做但不敢做”的阶段。",{"title":637,"description":638,"icon":639},"本地化需求","不同团队面向的语言、业务习惯、部署环境和组织流程都不一样。我们不是简单复制通用产品，而是把模型能力、本地业务流程和已有系统接起来，让方案真正适配你的团队，而不是要求团队反过来适应工具。","MapPinned",{"title":641,"description":642,"icon":643},"安全审计需求","只让模型“自觉遵守规则”是不够的，真正可上线的系统必须考虑日志留痕、权限控制、调用审计和风险隔离。我们更重视工程化安全边界，帮助客户建立可追踪、可解释、可审查的 AI 使用链路。","LockKeyhole",{"title":645,"description":646,"icon":647},"\"翻译\"需求","很多跨语言场景不是把内容翻出来就结束了，还涉及术语统一、语气控制、上下文保持和多语言协作效率。我们可以把翻译能力和知识库、工作流、审核机制结合起来，做成真正服务业务的语言系统。","Languages",{"title":649,"description":650,"icon":651},"成本需求","模型越强不代表越适合所有请求。我们会从路由策略、调用频率、任务分层、缓存和交付形态上一起设计，帮助客户把成本花在真正有价值的地方，而不是用最高配置去处理所有问题。","BadgeDollarSign","我们在做什么",[654,658,661,664,667],{"title":655,"description":656,"icon":657},"AI 教学","培训与科普，帮助用户理解与运用 AI。","GraduationCap",{"title":659,"description":660,"icon":95},"AIGC 与 GPT","整合生成式能力，提供更智能的支持。",{"title":662,"description":663,"icon":159},"企业级 AI API","定制化 API，满足企业与产品集成需求。",{"title":665,"description":666,"icon":170},"知识库搭建","构建可检索、可维护的知识体系。",{"title":668,"description":669,"icon":670},"AI 软件研发","定制开发与交付 AI 相关软件。","Code2","愿景",[673,674],"无论你要提效、解决问题还是创造新价值，开发喵 AI 都希望成为可靠助手。","我们努力降低技术门槛，让 AI 服务于生活与工作，并与合作伙伴建立长期可迭代的关系。",{"headTitle":676,"hero":677,"categoriesTitle":684,"categories":685},"我们的产品",{"eyebrow":678,"title":676,"description":679,"primary":680,"secondary":682},"Products","从 AI 应用开发到公共服务平台、知识库、训练和聚合能力，围绕实际交付组织产品矩阵。",{"label":681,"to":50},"联系合作",{"label":683,"to":38},"返回首页","产品分类",{"enterprise":686,"personal":689,"self":692,"global":695},{"title":687,"description":688},"企业服务","面向企业系统、交付项目和业务协同场景的服务能力。",{"title":690,"description":691},"个人服务","围绕账号、设备、数字内容和个人使用场景提供的服务入口。",{"title":693,"description":694},"自研产品","团队自主打磨的产品、平台和工程模板。",{"title":696,"description":697},"全球用户","面向跨区域访问、全球模型接入与国际化业务需求的产品能力。",{"headTitle":699,"hero":700,"introTitle":699,"intro":707,"valuesTitle":710,"values":711,"cooperationTitle":728,"cooperation":729},"团队介绍",{"eyebrow":701,"title":702,"description":703,"primary":704,"secondary":706},"Team","开发喵团队","围绕 AI 产品、开发工具链与业务落地协作，强调长期主义与工程质量。",{"label":705,"to":50},"访问开发者中心",{"label":25,"to":41},[708,709],"团队由一群长期关注 AI 产品、工程效率与交付质量的成员组成。","支持远程协作，重视主动沟通、明确边界与责任交付。","价值观",[712,716,720,724],{"title":713,"description":714,"icon":715},"主动学习","持续关注模型、工具链与交付方式的变化。","Lightbulb",{"title":717,"description":718,"icon":719},"交付负责","重视需求澄清、过程沟通与结果稳定性。","BadgeCheck",{"title":721,"description":722,"icon":723},"长期合作","倾向与客户建立长期、可迭代的协作关系。","Handshake",{"title":725,"description":726,"icon":727},"用户价值","把复杂技术转成真正可用的产品体验。","Heart","能力与合作方式",[730,731,732,733],"定制化开发与交付","AI 接入咨询与方案设计","知识库、RAG、Agent 模板沉淀","持续维护与版本迭代",{"headTitle":81,"hero":735,"collaborationTitle":744,"collaboration":745,"roadmapTitle":768,"roadmap":769,"joinTitle":774,"joinText":775,"githubCta":776,"sponsorTitle":777,"sponsorText":778,"sponsorCta":779},{"eyebrow":736,"title":81,"description":737,"primary":738,"secondary":741},"Developer Hub","面向 API 接入、功能共建、定制开发和技术合作，整理协作方式、计划与需求提交入口。",{"label":739,"to":740,"external":58},"邮件提交需求","mailto:hello@kaifamiao.com?subject=%E5%BC%80%E5%8F%91%E9%9C%80%E6%B1%82%E6%8F%90%E4%BA%A4",{"label":742,"to":743},"查看联系方式","\u002Flink","协作方式",[746,750,754,758,761,764],{"title":747,"description":748,"icon":749},"开放协作","欢迎面向产品、工具链、内容和基础设施共同建设。","Globe2",{"title":751,"description":752,"icon":753},"异步优先","支持远程协作，强调信息留痕与明确边界。","Clock3",{"title":755,"description":756,"icon":757},"模块化沉淀","偏好可复用组件、模板与基础设施积累。","Layers3",{"title":717,"description":759,"icon":760},"重视版本、质量、验收和后续维护。","CheckCheck",{"title":762,"description":763,"icon":201},"安全意识","涉及模型、数据和权限时强调审计与合规。",{"title":765,"description":766,"icon":767},"投资者","欢迎关注 AI 产品、基础设施与长期能力建设的投资者沟通交流，我们更看重长期主义、清晰路径和可持续增长。","LineChart","正在开发的计划",[770,771,772,773],"公共服务平台能力扩展与开放文档完善","多模型路由策略优化与成本观测看板","RAG \u002F Agent 工程模板的标准化组件沉淀","演示体系与教学资料持续迭代","加入我们","我们欢迎对 AI 产品、开发工具链、工程效率有热情的伙伴加入，重视主动沟通与责任交付。","访问 GitHub 团队主页","请喝咖啡","如果你认可我们的方向，愿意请我们喝杯咖啡，支持团队持续打磨产品、内容和基础设施，这里会保留一个直接支持入口。二维码放上来后，可以直接扫码支持。","也欢迎投资合作",{"headTitle":781,"hero":782,"narrativeTitle":652,"narrative":790,"focusTitle":793,"focus":794,"whyTitle":801,"why":802,"contactTitle":806,"contactText":807},"投资者关系",{"eyebrow":783,"title":781,"description":784,"primary":785,"secondary":788},"Investor Relations","面向关注 AI 产品、知识系统、基础设施与长期能力建设的投资者，集中说明我们的方向、阶段、合作方式与沟通入口。",{"label":786,"to":787,"external":58},"邮件联系","mailto:hello@kaifamiao.com?subject=%E6%8A%95%E8%B5%84%E5%90%88%E4%BD%9C%E6%B2%9F%E9%80%9A",{"label":789,"to":50},"返回开发者中心",[791,792],"我们围绕 AI 应用交付、知识库、模型接入、原理演示和公共服务能力，持续沉淀可复用产品与工程资产。","重点不是做短期概念包装，而是做能上线、能维护、能扩展、能形成长期积累的 AI 产品体系。","当前关注方向",[795,796,797,798,799,800],"企业级 AI 应用交付","知识库与 RAG 产品化","多模型接入与路由层","教学演示与认知型产品","公共服务平台能力沉淀","长期可维护的 AI 工程模板","为什么值得沟通",[803,804,805],"方向围绕真实交付与长期产品资产，而不是短期热点","产品、演示、知识系统和服务能力之间能够互相支撑","适合长期主义、偏工程和基础能力建设的合作逻辑","沟通方式","如果你关注 AI 产品、基础设施、知识系统和可持续增长，可以直接通过邮件联系。我们更希望以务实、透明和长期视角展开交流。",{"headTitle":809,"hero":810,"categoriesTitle":817,"categoriesDescription":818,"categories":819,"conceptsTitle":821,"conceptsDescription":832,"openLabel":833,"nextTitle":834,"nextDescription":835},"AI原理演示",{"eyebrow":811,"title":809,"description":812,"primary":813,"secondary":816},"Demos","把大模型、RAG 与向量空间的核心处理链路拆解成更容易观察和理解的可视化页面。",{"label":814,"to":815},"打开 LLM Demo","\u002Fdemo\u002Fllm",{"label":676,"to":41},"学习路径","按主题理解这些演示，会比按时间顺序逐个点开更容易形成整体认识。",{"fundamentals":820,"retrieval":823,"orchestration":826,"platform":829},{"title":821,"description":822},"基础概念","先建立对 token、上下文、提示词和采样的直觉，再看工程演示会更顺。",{"title":824,"description":825},"RAG \u002F 检索","理解向量、召回、分块、重排和阈值，才能看懂知识库为什么有时答得准、有时会偏。",{"title":827,"description":828},"Agent \u002F 编排","关注技能路由、工具调用、结构化输出和安全边界，理解模型如何接近真实系统。",{"title":830,"description":831},"工程平台","看平台如何处理模型路由、资源分配和质量成本平衡。","先理解这些概念，再去看动画，会更容易把模型处理过程和工程限制对应起来。","打开演示","推荐下一步","继续沿着相关主题看下去，会更容易把概念和工程实现串起来。",{"headTitle":78,"hero":837,"stepsTitle":844,"steps":845,"servicesTitle":850},{"eyebrow":838,"title":78,"description":839,"primary":840,"secondary":842},"Key Access","统一整理短信、大模型、网络与 IP 检测服务，便于研发团队集中查看开通流程、Key 使用方式与服务明细。",{"label":841,"to":160},"返回 AI API 服务",{"label":843,"to":41},"查看全部产品","开通流程",[846,847,848,849],"提交需求与业务场景","获取评估与接入建议","开通 Key 与权限","对接文档并进入联调","四项公共服务",{"headTitle":852,"hero":853},"AI 喵帮助",{"eyebrow":854,"title":852,"description":855,"primary":856,"secondary":859},"Help Center","文档、常见问题、接入说明与使用指引已统一整理在帮助中心，建议优先从帮助中心入口查看最新内容。",{"label":857,"to":858,"external":58},"进入帮助中心","https:\u002F\u002Fhelo.1155665ai.com",{"label":683,"to":38},{"headTitle":861,"hero":862,"openLabel":867},"相关链接",{"eyebrow":863,"title":861,"description":864,"primary":865,"secondary":866},"Links","团队主页、服务入口与联系方式集中在这里，方便快速跳转。",{"label":683,"to":38},{"label":81,"to":50},"打开",[869,875,881,887,892,897,903,908,914,921,925,929,935,940,945],{"badge":870,"category":214,"title":871,"description":872,"icon":670,"to":873,"cta":874},"01","AI 应用软件开发","面向业务流程打造可交付 AI 应用，覆盖前端、后端与模型集成。","\u002Fproducts\u002Fai-application-development","查看详情",{"badge":876,"category":877,"title":878,"description":879,"icon":170,"to":880,"cta":874},"02","self","everyone-llm（知识库）","沉淀文档检索、问答和企业知识管理能力。","\u002Fproducts\u002Feveryone-llm",{"badge":882,"category":883,"title":884,"description":885,"icon":886},"03","global","海外家宽","提供稳定跨区域网络能力，支持业务拓展与测试场景。","Globe",{"badge":888,"category":883,"title":889,"description":890,"icon":159,"to":891,"cta":874},"04","全球大模型 API 🔥","统一多模型接入、鉴权与配额管理。","\u002Fproducts\u002Fglobal-llm-api",{"badge":893,"category":214,"title":894,"description":895,"icon":165,"to":896,"cta":874},"05","大模型训练","从数据、训练到评测与上线的全流程支持。","\u002Fproducts\u002Fllm-training",{"badge":898,"category":214,"title":899,"description":900,"icon":901,"to":902,"cta":874},"06","AI 产品聚合服务","聚合文本、图像、语音、视频等多类 AI 能力。","Layers","\u002Fproducts\u002Fai-product-aggregation",{"badge":904,"category":214,"title":905,"description":906,"icon":907},"07","自研 DeepSeek 本地服务器","提供面向企业内部部署、推理调用和模型运行环境管理的 DeepSeek 服务器支持。","ServerCog",{"badge":909,"category":877,"title":910,"description":911,"icon":912,"to":913,"cta":874},"08","AI 多模型决策平台 OpenLens","多模型对话决策系统，支持比较、对话与策略辅助。","ScanEye","\u002Fproducts\u002Fopenlens",{"badge":915,"category":916,"title":917,"description":918,"icon":919,"to":61,"external":58,"cta":920},"09","personal","手边小卖部 HandyHub","账号与数字商品服务入口。","Store","访问官网",{"badge":922,"category":916,"title":923,"description":924,"icon":170},"10","亚马逊 Kindle 服务","围绕 Kindle 账号、设备与内容使用场景提供配套服务支持。",{"badge":926,"category":916,"title":927,"description":928,"icon":170},"11","全球电子书AI增强","面向全球电子书阅读、整理、问答与辅助理解场景的 AI 服务能力。",{"badge":930,"category":877,"title":931,"description":932,"icon":933,"to":934,"cta":874},"12","kfm_nuxtchat","面向业务系统的 AI 聊天与工作流集成模板。","MessageSquareCode","\u002Fproducts\u002Fkfm-nuxtchat",{"badge":936,"category":214,"title":937,"description":938,"icon":939,"to":79,"cta":141},"13","开发喵公共服务平台","统一提供短信、模型、网络与 IP 检测服务。","KeyRound",{"badge":941,"category":214,"title":942,"description":943,"icon":907,"to":944,"cta":874},"14","自主式 AI 部署 🔥","面向 Agent、RAG 与多服务协同的工程化部署方案。","\u002Fproducts\u002Fautonomous-ai-deployment",{"badge":946,"category":883,"title":947,"description":948,"icon":767,"to":949,"cta":874},"15","全球金融市场数据接口 🔥","提供高频与结构化市场数据接入能力。","\u002Fproducts\u002Fglobal-financial-market-data-api",[951,954,957],{"title":952,"description":953,"icon":670},"什么是 Token","Token 是模型处理文本时使用的最小计量单位，不完全等于“一个汉字”或“一个英文单词”。模型会先把输入拆成 token，再进行编码、注意力计算和生成。英文里常常一个词会拆成多个 token，中文里短句也可能按词块或子词切分。理解 token，有助于理解计费、速度和上下文限制。 ",{"title":955,"description":956,"icon":901},"LLM 上下文长度怎么理解","上下文长度可以理解为模型当前一次能“看见”的总 token 数量，里面同时包含系统提示词、历史对话、用户新输入以及模型生成内容。它不是只算你的提问文本。上下文越长，模型能参考的信息越多，但成本、延迟和注意力分配压力也会增加。",{"title":958,"description":959,"icon":95},"什么是 Skills","Skills 可以理解为给模型预先准备好的能力模块或工作说明。它不只是知识说明，更像一套可复用的操作规范，告诉模型在某类任务里该怎样判断、调用什么工具、遵循什么输出格式。Skills 做得好，模型表现会更稳定，也更接近真正可交付的系统。",{"ai-application-development":961,"llm-training":990,"ai-product-aggregation":1018,"everyone-llm":1046,"openlens":1073,"autonomous-ai-deployment":1100,"global-financial-market-data-api":1128,"global-llm-api":1156,"kfm-nuxtchat":1184},{"id":204,"slug":962,"title":963,"eyebrow":964,"summary":965,"icon":670,"intro":966,"features":969,"scenarios":976,"advantages":983,"primaryCta":987,"secondaryCta":989},"ai-application-development","AI应用软件开发","AI Applications","围绕真实业务流程设计和交付 AI 应用，把模型能力接入前端、后端、工作流与运维体系。",[967,968],"这类项目的重点不是把模型接上去，而是把角色、流程、输入输出、权限和稳定性一起设计清楚。","我们更关注交付后的可维护性，包括接口边界、日志观测、错误处理、版本迭代和业务方可用性。",[970,971,972,973,974,975],"业务流程梳理","前后端一体化交付","模型接入与工作流设计","权限与日志机制","多语言界面支持","上线与持续迭代",[977,978,979,980,981,982],"企业内部 Copilot","智能客服与售后支持","内容生产平台","运营分析助手","审批和知识协同系统","跨系统工作流集成",[984,985,986],"强调业务落地而不是单点 Demo","从产品体验到工程交付一体化推进","支持后续版本持续迭代与能力扩展",{"label":988,"to":50},"联系开发者中心",{"label":25,"to":41},{"id":244,"slug":991,"title":894,"eyebrow":992,"summary":993,"icon":165,"intro":994,"features":997,"scenarios":1004,"advantages":1011,"primaryCta":1015,"secondaryCta":1016},"llm-training","Model Training","覆盖数据整理、训练、评测、推理优化和上线交付的大模型训练服务。",[995,996],"不是所有场景都需要从零训练，但很多场景需要结合业务数据做微调、对齐和专项评测。","我们会根据预算、算力和目标效果，在训练强度、部署方案和上线节奏之间做实际权衡。",[998,999,1000,1001,1002,1003],"训练数据清洗与构建","指令微调方案设计","评测指标与回归验证","推理性能优化","GPU 与资源规划","上线和运维方案",[1005,1006,1007,1008,1009,1010],"行业垂直模型","企业知识问答增强","内容生产能力强化","分类与审核模型","专项语义理解任务","定制化对话系统",[1012,1013,1014],"兼顾实验效率与交付质量","支持性能、成本和效果三方平衡","可与知识库、RAG 和 Agent 体系联动",{"label":681,"to":50},{"label":1017,"to":41},"返回产品页",{"id":281,"slug":1019,"title":1020,"eyebrow":1021,"summary":1022,"icon":901,"intro":1023,"features":1026,"scenarios":1033,"advantages":1040,"primaryCta":1044,"secondaryCta":1045},"ai-product-aggregation","AI产品聚合服务","AI Aggregation","统一接入文本、图像、语音、视频等多类 AI 能力，提供聚合层、路由层和运营层支撑。",[1024,1025],"当产品需要同时使用多种模型能力时，聚合层决定了接入成本、扩展速度和后续维护复杂度。","我们提供的不只是模型列表，而是统一鉴权、调用封装、路由策略和能力编排。",[1027,1028,1029,1030,1031,1032],"多能力统一接入","调用封装与鉴权","模型与能力路由","错误兜底机制","成本观测与配额管理","跨模型组合编排",[1034,1035,1036,1037,1038,1039],"AI 工具平台","企业统一 AI 能力层","多模态产品接入","内部研发平台","SaaS 能力聚合","对外 API 中台",[1041,1042,1043],"降低多模型接入复杂度","让产品团队更快试错和扩展","适合做长期能力中台和服务层",{"label":625,"to":50},{"label":1017,"to":41},{"id":318,"slug":1047,"title":878,"eyebrow":1048,"summary":1049,"icon":170,"intro":1050,"features":1053,"scenarios":1060,"advantages":1067,"primaryCta":1071,"secondaryCta":1072},"everyone-llm","Knowledge Base","面向团队知识沉淀、检索问答和内容协同的知识库产品，把文档、规则和业务经验整理成可被检索和复用的工作资产。",[1051,1052],"知识库的价值不只是“能搜到”，而是让团队能把分散的文档、流程、FAQ 和经验真正转成可引用、可维护、可更新的知识系统。","everyone-llm 更适合用在企业内部知识管理、客户支持、 SOP 沉淀和多角色协同场景中，让问答和检索真正服务业务。",[1054,1055,1056,1057,1058,1059],"文档接入与整理","RAG 检索问答","多知识源管理","引用片段与答案溯源","知识更新与维护","团队协同使用",[1061,1062,1063,1064,1065,1066],"企业内部知识库","售前售后问答","SOP 与操作手册查询","项目资料沉淀","培训与上手支持","跨团队经验共享",[1068,1069,1070],"把知识沉淀和问答能力放到同一产品里","更适合长期维护而不是一次性导入","能与 RAG、Agent 和工作流体系继续联动",{"label":625,"to":50},{"label":1017,"to":41},{"id":355,"slug":1074,"title":910,"eyebrow":1075,"summary":1076,"icon":912,"intro":1077,"features":1080,"scenarios":1087,"advantages":1094,"primaryCta":1098,"secondaryCta":1099},"openlens","Multi-Model Platform","面向多模型比较、问答评估和决策辅助的产品平台，帮助团队更高效地理解不同模型的差异与适用场景。",[1078,1079],"OpenLens 不是简单把多个模型放在一个页面里，而是把比较、观察、试用和策略判断放到同一工作界面里。","它更适合需要做模型选型、回答对比、质量观察和策略试验的个人用户与团队。",[1081,1082,1083,1084,1085,1086],"多模型并行对话","回答差异对比","策略与效果观察","统一交互入口","多场景试验支持","辅助模型选型",[1088,1089,1090,1091,1092,1093],"模型选型对比","Prompt 测试","回答质量观察","内部评估与演示","策略辅助决策","多模型工作流入口",[1095,1096,1097],"更适合做横向比较与判断","降低多模型试验成本","把观察和决策放在同一界面里",{"label":920,"to":57,"external":58},{"label":1017,"to":41},{"id":392,"slug":1101,"title":1102,"eyebrow":1103,"summary":1104,"icon":907,"intro":1105,"features":1108,"scenarios":1115,"advantages":1122,"primaryCta":1126,"secondaryCta":1127},"autonomous-ai-deployment","自主式AI部署","AI Deployment","面向 Agent、RAG、多模型调用和私有环境的工程化部署方案，强调稳定、可观测和可扩展。",[1106,1107],"部署不是把服务启动起来就结束，还需要处理资源规划、权限边界、日志观测、版本升级和异常回滚。","我们更关注可长期运行的工程体系，而不是一次性的环境搭建。",[1109,1110,1111,1112,1113,1114],"私有化与混合部署","多服务编排","权限与访问控制","日志监控与告警","版本升级与回滚","资源与成本规划",[1116,1117,1118,1119,1120,1121],"企业内网部署","Agent 系统落地","知识库与 RAG 平台","私有模型服务","多模型路由中台","长期运行的 AI 基础设施",[1123,1124,1125],"强调上线后的可持续运维","适合复杂 AI 工程体系整合","兼顾安全、性能和可扩展性",{"label":988,"to":50},{"label":1017,"to":41},{"id":429,"slug":1129,"title":1130,"eyebrow":1131,"summary":1132,"icon":767,"intro":1133,"features":1136,"scenarios":1143,"advantages":1150,"primaryCta":1154,"secondaryCta":1155},"global-financial-market-data-api","全球金融市场数据接口","Market Data","提供面向全球市场的结构化金融数据接口，支持量化研究、监控、策略和产品接入。",[1134,1135],"数据接口的价值不只是拿到数据，而是拿到稳定、结构清晰、适合下游系统消费的数据形态。","我们会更关注字段设计、调用稳定性、频率限制、观测能力和后续扩展。",[1137,1138,1139,1140,1141,1142],"全球市场数据接入","结构化字段设计","稳定调用与限频控制","历史与实时数据支持","接口文档与接入辅助","监控与异常观测",[1144,1145,1146,1147,1148,1149],"量化研究","交易监控看板","金融资讯产品","分析助手","策略信号系统","跨市场数据聚合",[1151,1152,1153],"接口更贴近下游应用消费","支持全球市场与扩展需求","适合产品化和工程化接入",{"label":681,"to":50},{"label":1017,"to":41},{"id":464,"slug":1157,"title":1158,"eyebrow":1159,"summary":1160,"icon":159,"intro":1161,"features":1164,"scenarios":1170,"advantages":1177,"primaryCta":1181,"secondaryCta":1183},"global-llm-api","全球大模型 API","Global LLM API","统一接入全球主流大模型能力，帮助产品和企业系统快速获得稳定的推理与生成接口。",[1162,1163],"对于需要面向全球用户和多模型策略的产品来说，统一 API 层可以明显降低接入和维护成本。","我们会把鉴权、路由、限流、日志、配额和成本控制放进同一层能力里。",[437,1165,1166,1167,1168,1169],"鉴权与配额管理","路由与限流机制","调用日志与监控","多语言接入支持","成本与使用观测",[1171,1172,1173,1174,1175,1176],"国际化产品接入","企业 AI 能力中台","内容生成平台","多模型实验和比较","客服与问答产品","开发者平台",[1178,1179,1180],"更适合全球化和多模型场景","统一接入层降低维护复杂度","支持后续扩展更多模型和能力",{"label":1182,"to":79},"进入公共服务平台",{"label":1017,"to":41},{"id":500,"slug":1185,"title":931,"eyebrow":1186,"summary":1187,"icon":933,"intro":1188,"features":1191,"scenarios":1198,"advantages":1205,"primaryCta":1209,"secondaryCta":1210},"kfm-nuxtchat","Chat Template","面向业务系统的 AI 聊天与工作流集成模板，适合快速构建带权限、上下文和业务动作的对话入口。",[1189,1190],"很多业务系统都需要一个能接模型、接流程、接权限的聊天入口，而不是单独的聊天窗口。kfm_nuxtchat 更适合做这类可继续扩展的集成模板。","它强调的是页面结构、会话管理、业务动作挂接和后续定制能力，适合作为企业内外部 AI 对话入口的起点。",[1192,1193,1194,1195,1196,1197],"聊天界面模板","上下文与会话管理","业务动作集成","权限与角色适配","多模型接入扩展","适合二次开发",[1199,1200,1201,1202,1203,1204],"企业内部 AI 助手","业务系统对话入口","客服与咨询窗口","流程型问答页面","知识问答前端","定制化 AI 产品原型",[1206,1207,1208],"不是孤立聊天页，而是业务集成模板","适合快速落地并继续扩展","能够承接权限、知识库和工作流需求",{"label":988,"to":50},{"label":1017,"to":41},[1212,1227,1242,1256,1269,1281,1293,1305,1317,1329,1341,1354,1367,1380,1392,1405,1418,1430,1443,1456,1469,1481,1493,1506,1518],{"id":204,"slug":1213,"category":1214,"title":1215,"summary":1216,"focus":1217,"accent":1218,"code":1219,"embedPath":1220,"steps":1221},"llm","fundamentals","LLM 问答过程动画 🔥","把提问、编码、推理、解码和输出的链路拆成可观察节点。","重点：Transformer 处理链路","from-blue-500 via-cyan-500 to-emerald-400","Tokenize -> Embed -> Attend -> Decode -> Answer","\u002Finteractive-demos\u002Fllm.html",[1222,1223,1224,1225,1226],"用户提问","Token 编码","上下文注意力","解码生成","答案输出",{"id":244,"slug":1228,"category":1229,"title":1230,"summary":1231,"focus":1232,"accent":1233,"code":1234,"embedPath":1235,"steps":1236},"rag","retrieval","RAG 检索增强生成 🔥","演示查询改写、召回、重排、拼接上下文与最终生成的完整流程。","重点：检索链路","from-sky-500 via-cyan-500 to-teal-400","Query -> Retrieve -> Rerank -> Context -> Generate","\u002Finteractive-demos\u002Frag.html",[1237,1238,1239,1240,1241],"用户查询","向量召回","重排筛选","上下文拼接","生成回答",{"id":281,"slug":1243,"category":1229,"title":1244,"summary":1245,"focus":1246,"accent":1247,"code":1248,"embedPath":1249,"steps":1250},"embedding","Embedding 向量空间","通过二维示意和相似度说明文本如何落入向量空间。","重点：向量空间直觉","from-amber-500 via-orange-500 to-rose-400","Text -> Vector -> Similarity -> Clusters","\u002Finteractive-demos\u002Fembedding.html",[1251,1252,1253,1254,1255],"原始文本","分词切块","向量映射","相似度计算","聚类结果",{"id":318,"slug":1257,"category":1214,"title":1258,"summary":1259,"focus":1260,"accent":1261,"code":1262,"embedPath":1263,"steps":1264},"token","什么是 Token 🔥","用动画把一句话拆成模型真正处理的 token，理解 token 不是“一个字=一个 token”。","重点：切分、计量与生成单位","from-fuchsia-500 via-violet-500 to-sky-400","Input -> Tokenize -> Count -> Process",null,[1265,1266,1267,1268],"输入文本","子词切分","token 计数","进入模型处理",{"id":355,"slug":1270,"category":1214,"title":1271,"summary":1272,"focus":1273,"accent":1274,"code":1275,"embedPath":1263,"steps":1276},"context-window","LLM 上下文长度","通过滑动窗口展示模型一次真正能“看到”的 token 范围，以及为什么旧内容会被截断。","重点：可见窗口与截断直觉","from-emerald-500 via-teal-500 to-cyan-400","System + History + User + Output \u003C= Context Window",[1277,1278,1279,1280],"系统提示词","历史对话","最新输入","生成输出",{"id":392,"slug":1282,"category":1283,"title":958,"summary":1284,"focus":1285,"accent":1286,"code":1287,"embedPath":1263,"steps":1288},"skills","orchestration","把 Skills 理解成给模型的能力模块，演示请求如何被技能路由并转成稳定执行过程。","重点：能力路由与执行规范","from-rose-500 via-orange-500 to-amber-400","Task -> Skill Match -> Tool Plan -> Structured Output",[1289,1290,1291,1292],"识别任务","匹配技能","生成执行计划","结构化输出",{"id":429,"slug":1294,"category":1214,"title":1295,"summary":1296,"focus":1297,"accent":1298,"code":1299,"embedPath":1263,"steps":1300},"prompt-structure","Prompt 结构演示","展示 system、user、assistant 示例如何被拼成最终输入，理解“提示词”不是单独一句话。","重点：消息结构与角色分工","from-indigo-500 via-sky-500 to-cyan-400","System + Few-shot + User -> Final Prompt",[1301,1302,1303,1304],"系统指令","示例消息","用户输入","拼接成最终提示",{"id":464,"slug":1306,"category":1214,"title":1307,"summary":1308,"focus":1309,"accent":1310,"code":1311,"embedPath":1263,"steps":1312},"transformer","Transformer 原理演示 🔥","用可视化方式展示 token 如何彼此关注、加权汇聚并形成新的上下文表示。","重点：自注意力与上下文建模","from-violet-500 via-indigo-500 to-cyan-400","Tokens -> Attention Scores -> Weighted Sum -> Contextual Output",[1313,1314,1315,1316],"输入 token","计算相关性","归一化权重","生成上下文表示",{"id":500,"slug":1318,"category":1214,"title":1319,"summary":1320,"focus":1321,"accent":1322,"code":1323,"embedPath":1263,"steps":1324},"temperature","Temperature 温度演示","用同一个问题对比低温和高温采样，理解模型为什么会更稳或更发散。","重点：随机性与稳定性","from-blue-500 via-violet-500 to-pink-400","Low Temp -> Stable | High Temp -> Diverse",[1325,1326,1327,1328],"同一提示词","设置温度","候选概率变化","输出风格差异",{"id":536,"slug":1330,"category":1229,"title":1331,"summary":1332,"focus":1333,"accent":1334,"code":1335,"embedPath":1263,"steps":1336},"rag-chunking","RAG 分块 Chunking 演示","对比大块、适中、小块切分对召回命中的影响，理解为什么 chunk 大小会改变答案质量。","重点：切块粒度与召回质量","from-emerald-500 via-teal-500 to-lime-400","Document -> Chunk -> Embed -> Retrieve",[1337,1338,1339,1340],"原始文档","不同粒度切块","向量化","召回命中差异",{"id":1342,"slug":1343,"category":1283,"title":1344,"summary":1345,"focus":1346,"accent":1347,"code":1348,"embedPath":1263,"steps":1349},11,"agent-tools","Agent 工具调用演示","展示 Agent 如何理解任务、挑选工具、读取结果并决定下一步，而不是一次性给答案。","重点：工具使用闭环","from-orange-500 via-amber-500 to-yellow-400","Task -> Tool -> Result -> Next Action",[1350,1351,1352,1353],"理解任务","选择工具","读取结果","继续决策",{"id":1355,"slug":1356,"category":1283,"title":1357,"summary":1358,"focus":1359,"accent":1360,"code":1361,"embedPath":1263,"steps":1362},12,"function-calling","Function Calling \u002F JSON 输出","展示模型如何把自然语言请求转成结构化参数，而不是只返回一段描述文字。","重点：结构化输出与参数映射","from-cyan-500 via-sky-500 to-indigo-400","Prompt -> Schema Match -> JSON Arguments",[1363,1364,1365,1366],"理解意图","匹配字段","生成 JSON","调用函数",{"id":1368,"slug":1369,"category":1214,"title":1370,"summary":1371,"focus":1372,"accent":1373,"code":1374,"embedPath":1263,"steps":1375},13,"chat-memory","多轮对话记忆","演示历史消息如何逐轮进入上下文，以及为什么对话越长越需要摘要和裁剪。","重点：历史消息与上下文占用","from-teal-500 via-emerald-500 to-lime-400","History + Latest Input -> Context Window",[1376,1377,1378,1379],"历史累积","上下文占用","摘要压缩","继续回答",{"id":1381,"slug":1382,"category":1229,"title":1383,"summary":1384,"focus":1385,"accent":1386,"code":1387,"embedPath":1263,"steps":1388},14,"rag-rerank","RAG 重排 Rerank","展示召回结果为什么还要重排，以及最终真正送进模型的片段通常只有少数几条。","重点：召回不等于最终采用","from-emerald-500 via-cyan-500 to-sky-400","Retrieve -> Score -> Rerank -> Keep Top Results",[1389,1390,1239,1391],"初始召回","相关性评分","送入生成",{"id":1393,"slug":1394,"category":1283,"title":1395,"summary":1396,"focus":1397,"accent":1398,"code":1399,"embedPath":1263,"steps":1400},15,"prompt-injection","Prompt 注入 \u002F 安全边界","说明为什么 system 指令、权限隔离和工具边界不能只靠模型“自觉遵守”。","重点：安全约束与越权风险","from-rose-500 via-red-500 to-orange-400","System Rules > User Injection > Guardrails",[1401,1402,1403,1404],"系统规则","恶意输入","安全检查","拒绝或隔离",{"id":1406,"slug":1407,"category":1229,"title":1408,"summary":1409,"focus":1410,"accent":1411,"code":1412,"embedPath":1263,"steps":1413},16,"embedding-threshold","Embedding 相似度阈值 🔥","通过相似度阈值控制展示为什么“有点像”不等于应该被采纳。","重点：相似度阈值与误召回","from-amber-500 via-orange-500 to-red-400","Vector Similarity >= Threshold ?",[1414,1415,1416,1417],"向量比较","计算相似度","设置阈值","决定是否采用",{"id":1419,"slug":1420,"category":1229,"title":1421,"summary":1422,"focus":1423,"accent":1360,"code":1424,"embedPath":1263,"steps":1425},17,"vector-database","向量数据库原理与存储","展示文本如何被写入向量库、建立索引，并在近邻检索时返回最相关的记录。","重点：向量写入、索引结构与存储记录","Document -> Embedding -> Vector Index -> ANN Search -> Metadata Hit",[1426,1427,1428,1429],"文本入库","生成向量","建立索引","近邻检索",{"id":1431,"slug":1432,"category":365,"title":1433,"summary":1434,"focus":1435,"accent":1436,"code":1437,"embedPath":1263,"steps":1438},18,"llm-distillation","大模型蒸馏","演示 Teacher 模型如何把能力迁移到更小的 Student 模型，以换取更低成本和更快响应。","重点：能力迁移、成本压缩与效果平衡","from-violet-500 via-fuchsia-500 to-rose-400","Teacher Output -> Distill -> Student Model",[1439,1440,1441,1442],"教师模型生成","软标签学习","对齐训练","学生模型上线",{"id":1444,"slug":1445,"category":365,"title":1446,"summary":1447,"focus":1448,"accent":1449,"code":1450,"embedPath":1263,"steps":1451},19,"llm-fine-tuning","大模型微调","展示通用模型如何通过业务数据微调，逐步适应特定领域语气、术语和输出格式。","重点：任务对齐、参数更新与效果提升","from-emerald-500 via-cyan-500 to-blue-400","Base Model + Domain Data -> Fine-tune -> Specialized Model",[1452,1453,1454,1455],"准备数据集","设定训练目标","参数更新","验证效果",{"id":1457,"slug":1458,"category":365,"title":1459,"summary":1460,"focus":1461,"accent":1462,"code":1463,"embedPath":1263,"steps":1464},20,"lora-vs-full-finetune","LoRA \u002F 全量微调对比","对比 LoRA 和全量微调在显存占用、训练成本、上线灵活性和效果提升上的差异。","重点：参数更新范围与工程取舍","from-indigo-500 via-violet-500 to-fuchsia-400","Base Model -> LoRA Adapters | Full Parameter Update",[1465,1466,1467,1468],"选择训练方式","更新参数范围","观察资源消耗","比较上线效果",{"id":1470,"slug":1471,"category":365,"title":1472,"summary":1473,"focus":1474,"accent":1475,"code":1476,"embedPath":1263,"steps":1477},21,"model-routing","模型路由","展示同一个请求为什么会按成本、速度和质量要求被分发给不同模型。","重点：路由策略与成本质量平衡","from-violet-500 via-indigo-500 to-sky-400","Task -> Route Policy -> Best Model",[1289,1478,1479,1480],"评估约束","分发模型","返回结果",{"id":1482,"slug":1483,"category":1229,"title":1484,"summary":1485,"focus":1486,"accent":1487,"code":1488,"embedPath":1263,"steps":1489},22,"rag-citations","RAG 引用来源","展示答案为什么需要带出处、片段编号和引用范围，帮助用户判断内容可信度。","重点：答案可信度与来源追踪","from-cyan-500 via-teal-500 to-emerald-400","Question -> Retrieve -> Answer + Citations",[1490,1491,1241,1492],"提出问题","召回片段","标注引用",{"id":1494,"slug":1495,"category":365,"title":1496,"summary":1497,"focus":1498,"accent":1499,"code":1500,"embedPath":1263,"steps":1501},23,"hallucination-check","模型幻觉与校验","通过对比“直接回答”和“经过校验的回答”，直观看到为什么系统需要事实校验与置信控制。","重点：幻觉风险与事实校验","from-pink-500 via-rose-500 to-orange-400","Draft Answer -> Verify -> Confidence Gate",[1502,1503,1504,1505],"生成草稿","检查事实","评估置信","输出结果",{"id":1507,"slug":1508,"category":1214,"title":1509,"summary":1510,"focus":1511,"accent":1512,"code":1513,"embedPath":1263,"steps":1514},24,"memory-summary","多轮对话摘要压缩","演示长对话为什么不能原样一直保留，以及系统如何把历史压成可复用摘要。","重点：记忆压缩与上下文节省","from-teal-500 via-cyan-500 to-sky-400","Long History -> Summary Memory -> Continue",[1376,1515,1516,1517],"触发压缩","生成摘要","继续对话",{"id":1519,"slug":1520,"category":1283,"title":1521,"summary":1522,"focus":1523,"accent":1411,"code":1524,"embedPath":1263,"steps":1525},25,"permission-routing","权限分层调用","展示同一个请求在普通用户、审核员和管理员权限下，会进入完全不同的工具链路。","重点：权限边界与工具分流","Request + Role -> Allowed Tools -> Result",[1526,1527,1528,1480],"识别身份","权限过滤","调用可用工具",{"ai-api":1530,"llm-development":1556,"ai-knowledge-base":1581,"rag-app":1608,"agent-app":1635,"blockchain":1661},{"id":204,"slug":1531,"title":157,"eyebrow":1532,"summary":1533,"icon":159,"intro":1534,"features":1537,"scenarios":1543,"advantages":1550,"primaryCta":1554,"secondaryCta":1555},"ai-api","AI Service","稳定的多模型接入与统一调用能力，帮助现有系统快速获得推理与生成能力。",[1535,1536],"支持文本、图像、语音等能力的统一接入，适合现有业务系统平滑升级。","提供鉴权、路由、限流和日志观测，降低业务团队的模型接入复杂度。",[1538,1539,1167,1540,1541,1542],"统一鉴权与计费","多模型路由","弹性扩缩容","多语言 SDK","成本透明",[1544,1545,1546,1547,1548,1549],"智能客服","内容生成","文本分析","语音与图像处理","翻译场景","推荐与个性化",[1551,1552,1553],"微服务与分布式架构","支持高并发与弹性扩缩","配套文档、SDK 与接入支持",{"label":1182,"to":79},{"label":25,"to":41},{"id":244,"slug":1557,"title":163,"eyebrow":1558,"summary":1559,"icon":165,"intro":1560,"features":1563,"scenarios":1570,"advantages":1575,"primaryCta":1579,"secondaryCta":1580},"llm-development","Model Engineering","覆盖数据治理、训练、评估、微调与上线的完整大模型工程服务。",[1561,1562],"适用于企业知识问答、行业模型适配和专项能力强化等场景。","从训练管线设计到推理部署，兼顾实验效率与工程可交付性。",[1564,1565,1566,1567,1568,1569],"训练数据清洗","指令数据构建","微调与评测","推理优化","GPU 训练经验","上线方案",[1005,977,1571,1572,1573,1574],"客服增强","内容审核","专用生成任务","多轮对话优化",[1576,1577,1578],"兼顾实验与工程落地","支持成本和性能权衡","可与知识库和 Agent 方案联动",{"label":681,"to":50},{"label":683,"to":38},{"id":281,"slug":1582,"title":168,"eyebrow":1048,"summary":1583,"icon":170,"intro":1584,"features":1587,"scenarios":1594,"advantages":1601,"primaryCta":1605,"secondaryCta":1607},"ai-knowledge-base","把分散文档沉淀为可检索、可维护、可审计的知识系统。",[1585,1586],"围绕文档治理、切片、索引、权限和问答效果做整体设计。","适合 SOP、售后手册、制度、产品资料与项目文档场景。",[1588,1589,1590,1591,1592,1593],"文档治理","结构化切片","向量索引","权限控制","知识更新流程","效果评估",[1595,1596,1597,1598,1599,1600],"企业内部知识问答","售前资料检索","客服辅助","研发文档检索","项目交付文档管理","培训助手",[1602,1603,1604],"强调知识生命周期管理","支持权限与可审计","可与 RAG 和 Agent 方案组合",{"label":1606,"to":176},"查看 RAG 服务",{"label":625,"to":50},{"id":318,"slug":1609,"title":173,"eyebrow":1610,"summary":1611,"icon":175,"intro":1612,"features":1615,"scenarios":1621,"advantages":1628,"primaryCta":1632,"secondaryCta":1634},"rag-app","Retrieval Augmented Generation","围绕召回、重排、上下文构造与生成效果优化的 RAG 工程服务。",[1613,1614],"适合需要把业务语料接入模型、提升回答准确性与可控性的场景。","关注可解释性、召回质量、权限和落地体验，而不只是把模型接上去。",[1616,1617,1618,1619,1620,1593],"查询改写","召回与重排","上下文治理","引用与出处","权限隔离",[1622,1623,1624,1625,1626,1627],"企业问答","帮助中心","售后支持","运营知识助手","法务检索","教育资料问答",[1629,1630,1631],"更贴近业务语料","比纯生成更可控","可与知识库和 API 体系联动",{"label":26,"to":1633},"\u002Fdemo\u002Frag",{"label":683,"to":38},{"id":355,"slug":1636,"title":178,"eyebrow":1637,"summary":1638,"icon":180,"intro":1639,"features":1642,"scenarios":1648,"advantages":1655,"primaryCta":1659,"secondaryCta":1660},"agent-app","AI Agent","构建具备任务拆解、工具调用、状态管理与执行回路的 Agent 系统。",[1640,1641],"适合流程自动化、复杂问答、跨系统执行和多步骤分析任务。","重点是把 Agent 从概念演示做成可观测、可回滚、可维护的工程系统。",[1643,1644,1645,1591,1646,1647],"任务编排","工具调用","状态持久化","失败重试","执行观测",[1649,1650,1651,1652,1653,1654],"运营自动化","智能分析助手","跨系统工单流转","内部流程执行","知识与工具结合问答","多角色协作",[1656,1657,1658],"强调工程稳定性","支持与现有业务系统集成","便于沉淀模板和组件",{"label":988,"to":50},{"label":25,"to":41},{"id":392,"slug":1662,"title":183,"eyebrow":1663,"summary":1664,"icon":100,"intro":1665,"features":1668,"scenarios":1675,"advantages":1682,"primaryCta":1686,"secondaryCta":1687},"blockchain","Blockchain","覆盖智能合约、链上交互、钱包接入与 DApp 交付的区块链开发服务。",[1666,1667],"支持链上产品从原型、合约设计到前后端接入与部署上线。","适合需要资产逻辑、链上凭证、交易流程或去中心化交互的业务。",[1669,1670,1671,1672,1673,1674],"智能合约开发","钱包接入","链上交易流程","DApp 前端","测试与审计协同","部署上线",[1676,1677,1678,1679,1680,1681],"链上资产管理","会员与凭证系统","链游功能模块","支付与结算场景","交易数据看板","链上身份场景",[1683,1684,1685],"前后端与链上协同交付","可与 AI 服务结合形成复合方案","支持定制化产品开发",{"label":625,"to":50},{"label":683,"to":38},{"source":1689},"directus",1777362836757]