在 1024 模力工场·AI 社区日主论坛上，上海鲸阙人工智能有限公司联合创始人，帆软软件战略副总裁沈涛，以《AI 助力企业找增长方向、找商机客户，实现效率到利润的转化》为题作主题分享。
沈涛提出了一个足以打动大多数企业老板的问题：AI 让员工一年多做 1000 份幻灯片，能多带来多少利润？
对此，分享了一次化工企业的交流：大家讨论了两个小时的研发、生产提效，休息时却发现，选定的技术路线可能已经失去竞争力，这让他更加关注一个问题：
方向错了，AI 可能只是让企业更快地走错路。
沈涛介绍，他们团队会用 AI 综合分析企业的招聘、专利、招投标和财务数据，从中寻找潜在商机。比如，一家企业招标采购 3150 千伏安的变压器，同时大量招聘检验、测试人员，再结合相关专利信息，就可以推测它可能正在扩大产能，并有加强质量控制的需求。对检测设备商、数据分析公司和人力服务公司来说，这就是值得进一步跟进的客户线索。
在一家制造企业探索电池管理系统业务时，团队则先梳理产业链、同行和下游市场，再筛选潜在客户及其近期需求，让“寻找新增长点”逐步落到具体客户上。他还介绍，AI 可以从活动报道等资料中提取人物与企业的关联，帮助寻找可能的商务引荐路径。
在沈涛看来，专业数据、业务知识和 AI 分析结合起来，才能更有针对性地辅助企业选方向、找客户、跟进商机。而提效最终有没有价值，还要看这些线索能否转化为订单、带来利润。
以下为沈涛讲演的完整速记，模力工场在不改变原意的前提下进行了订正和编辑。
员工用 AI 提效了，企业为什么没多赚钱？
首先我做个自我介绍，我之前一直在帆软，跟数据打了 10 年的交道。我们过去在 10 年的时间里面呢，其实我们发现一直给企业提供软件工具，提供效率工具，其实说实话，软件这个产品形态，是很难真正进入业务去给企业去解决增长问题和解决它的实际的业务问题的。而最多能够提供的呢，其实就是一个效率工具去帮他去解决这个效率问题。
而今年我从帆软出来开始创业呢，其实也就是我意识到 AI 的到来，确实它可以成为一个数字员工，甚至成为一些专家员工，去真正帮助去解决企业的利润问题和企业的增长问题。
那么今年呢，其实我观察了非常多的企业，尤其是虽然说我从 23 年开始研究 AI，但是说实话，正儿八经 AI 进入企业其实是从今年开始的。然后在今年开始的时候呢，其实我就观察到了一个非常大的现象，就是在现在 Workbody 也好，千问也好，还是豆包也好，这些个人级别的 AI 工具非常火热的时代呢，说实话，一家企业里面要去完成个人的 AI 提效是一件非常简单的事情。
但是的话，我跟很多的 1 号位还有老板去聊的时候，他其实就会很关心一个问题，就看似我 WorkBuddy 确实能够让我的员工做幻灯片的时间、做报告的时间变长变短了，但是一年时间你多帮我做 1000 份幻灯片，那又有什么用呢？对吧？他根本就是解决不了这个利润的问题的。
包括今年 3 月份的时候，当时我去聊了一家企业，他差不多是 400 亿的营收， 8000 人。今年二三月份 OpenClaw 火了之后呢，在企业内部大量的去铺这个 OpenClaw，在一个月时间差不多就花掉了他们 25% 的 AI 预算。但 AI 预算完了之后呢，他们在上面搭建了 60 多个智能体。 60 多个智能体呢，看似是欣欣向荣的，但实际上最后一算 账发现，诶，这些好像最终的 token 的输入就转变为了员工摸鱼的时间。
所以呢，其实我发现：个人的提效，它其实不一定等于组织的提效，而且即便你组织提效了，它也不一定等于利润的提升。
整一个企业 AI 落地，说实话它是一个非常大的一个系统工程。
就比如说这家企业，是之前我比较熟悉的一家企业，这家企业我名字我就不讲，但大家都非常熟悉。你看他们在过去 5 年时间里面，不管是数字化的基建，还是数字化的转型，还有 AI 的落地，其实都做的非常非常好的。他们就围绕整一个数字化的基建和数字化能力，还有 AI 的能力呢，是搭建了一套，就是围绕爆品增长的一套体系。
这套体系支撑他们 23 年到 25 年利润， 23 年到 25 年他的收入是翻了很多倍的，这个他们的这个组织效率和经营效率相比于同行来说是很好的。但是如果你看他的利润，你会发现他的收入是在往上走的，但是他的利润反而是每年都在往下跌的。所以呢，其实我观察到了这些现象之后呢，其实我就发现，就是我刚才提到的。就是在一家企业里面，你要完成 AI 在员工侧的落地，去提升个人的效率，再到个人效率到组织效率之间，以及组织的效率提升之后，你真正把这些效率转变为利润，这个其实非常困难的。
所以呢，其实我觉得在今年到三年内吧，企业里面的 AI 落地，有且只有一个主线叙事逻辑，就是你怎么用 AI 去提高效率，并且把这个效率变成利润。这个使命和这个问题呢，其实也是我创业的一个初心吧。我就在思考，有什么样的办法，以及有什么样的 AI 的能力，能够帮助企业把这个效率真正转变为利润的。
那么在回答这个命题之前呢，其实有一家我去拜访过的，也非常触动的一家公司吧，这个是浙江在丽水的那一家的公司，它差不多是年营收 60 个亿，是国内在化工领域做的非常强的。当时我们类似这样的研讨会的会场，然后去聊就是 AI 怎么帮助我做研发提效，做生产提效。他们的背景是呢，就是当时看到了半导体还有新能源这个行业的增长，对于一类新材料，就含氟类新材料的增长，他们看到了这个机会，并且做出了一个战略决策，说我要去研发一个含氟类的新产品。
我们才有了这个研讨会，说我这个 AI 能不能帮我去读论文，帮我去解构一些化学的逻辑？ AI 能不能帮我去提高质检的效率？ AI 能不能帮我去提高生产的效率？ AI 能不能帮我去提高设备巡检的这些效率？当时这个其实差不多聊了两个小时，聊了两个小时之后呢，确实做了很多关于 agent 的探讨，但是我们在聊了两个小时中场休息的时候，各位要知道一家大型企业真正聊核心话题，真正讲真话的时候，其大概率都不是在会议室，对吧？都在外面的抽烟处。
所以我们当时我们在会议，在那个中场休息在外面的时候聊呢，其实我就问了一个问题。就是你们现在已经看中了含氟类新材料方向的机会，并且我们再去讨论 AI 提效的一些事情。但是含氟类的新产产品方向到底是不是一个正确的机会方向？实际上中场休息的时候，大家都发现其实这个已经被市场淘汰了，已经被同行淘汰了，已经被他的客户给淘汰了。所以呢，就他们在一个已经被淘汰的技术路线上面去研讨很多关于 AI 提效的事情，我觉得是没有任何意义的。
所以这个其实就是我一直在思考的这个话题，如果说你在一个错误的方向上，你去聊很多 AI 提效的东西， AI 反而是会让你错的更加快的。
所以企业的效率到企业利润之间，其实有一个巨大的一个命题，就是你必须在一个对的方向上面，再去做对的事情，才能够把这个 AI 的效率变成利润，否则 AI 只会让你亏得亏得越多。因为这里面其实有一个公式的，就是一家企业的利润，它是什么？它就是你要在正确的时间做出正确的决策，这个其实就正确的方向，你在这个正确的方向上面，你再用正确的方法去提高你内部的执行效率。你才能够使得你的利润，才能够使得你这个效率变为这个利润。那么其实我会发现，其实现在 AI 来了之后，很多的 AI 厂商和 AI 的 provider，他都会提供，就是大量的数字员工。但是我认为那些数字员工其实都是在这个公式的右半部分，就是提高你的执行效率。那有没有一个 AI 工具能够帮助你去找到正确的时间？能不能帮你助你去找到正确的方向呢？这个我又带着过去 10 年跟数据打交道的经验呢，其实我就发现，就数据里面其实是蕴藏了很多值得被深究和值得被去研究的东西的。
所以呢，就是我现在在思考一个问题，就是，现在 AI 很强了， AI 也有长了很多的外部的大脑，它能够帮你去读材料，但是它能读到的材料，说实话都是公开的，它能读到百度上的东西，它能读到你公开网络搜索能搜索到的东西，但它有一些数据源是企业真正关心，但是 AI 现在其实是比较难读到的。比如说基础的工商信息，比如说 Wind 里面的上市公司财务的数据，比如说专业的招投标的数据，比如说类似 boss 直聘里面招聘的数据，又比如说企业跟企业风险相关的涉诉的数据，甚至说因为前段时间像常州那家公司，它也出现了一些舆情，对吧？那么类似这种舆情，它是在小红书上面，它是在抖音上面，它是在微信这些封闭生态上面去蕴藏的。你在豆包里面就问，你可能能问到跟抖音相关的东西，你在元宝里面问，你可能能问到跟微信相关的一些数据，但是你没有在一个平台上面能够把这些封闭生态的这个数据也能够问的清楚。而这些数据反而是我认为一家企业去判断一个方向，去找出一个对的客户，找到一个对的方向，才是真正需要的一些数据。
所以呢，其实就是我现在出来呢，就是汇聚了这一大批多元异构的，我认为对企业真正有专业价值的，以及这些你需要可能以前要在 N 个地方开 VIP 的，这些数据源我把它都整合起来。整合起来之后呢，其实有一件事情就变得比较有意思了，就比如说这个，就如果我们买了一批，就是招聘的这个数据库，那么在招聘这个数据库里面呢，其实会有很多有意思的现象，比如说这个是一家公司，它的这个 JD，你看它 JD，这个 JD 里面，财务总监他的职责是什么？职责是要牵投子公司上市，财务方面的统筹规划。你看这个数据，如果你把招聘这个数据，而不是通过爬虫的方式把它累积下来，你就会发现，诶，这家公司是要上市了。但你会发现上市的信号，其实对于另一家企业来说，它其实有帮助的，对吧？所以这个是一个，当你把这些数据都囊括在一起之后，很多维度的信息被 AI 就能看得见了。
第二个就是 AI 它因为有非常强的智能，它是能够看得懂一些数据源的。就比如说这个，如果我们有专业的招聘数据、专业的专利数据、专业的招投标数据，你看这个其实就比较有意思。你看这家公司在 26 年 6 月份的时候，他有一个厂房，高压电和负载满了，所以他还在招投标里面新增了一个 3150 千伏安的变压器的招标，这个是一个信号。另一个的话就是你会发现他在最近 3 周内急招检验员、测试员和耦验员。你看这两者数据如果真的能够被 AI 很好的消化的话，其实 AI 是能够得出一些判断的。
这个判断是什么？
第一个，哎，这家厂房已经不够了，所以他新增了变压器，说明什么？说明这家公司的产能在扩张。
第二个，它在产能扩张的过程中，因为他在招质检员，背后说明什么？说明他的良品率是在承压的。所以呢，就是 AI 去解读的时候，他可以发现这家公司产能在扩张，并且扩张过程中呢是跟光模块相关的耦合工序，这个工序呢他是会去申请一些专利作为保护的。
第三个，它在扩张过程中，这个产能瓶颈从产能变到了良品率，所以在大规模的招这些岗位。那么这个信号真正被 AI 解读出来之后，你会发现这个信号对谁是有用的？对于那些做厂房地产中介的来说，诶，他是有用的；因为他要做厂房扩建，我可以给他去服务。
对于帆软这种做数据分析的公司来说也是有用的，因为他的产能在扩张的过程中呢，他又需要有一些质检的需求、看板的需求、分数据分析的需求。那么对于那些做人力服务的公司来说，诶，他也是有需求的，因为他肯定要用工。对吧？那我可以给其他工人的服务，以及对于其他的一些做检测设备商来说，他也是有需求的。
所以呢，这些信号真正转变过来，转变为 AI 可被理解的一些信号和结论之后呢，对于另一家企业来说，他其实就是成为了一个商机了。所以呢，我发现就是数据跟智能结合在一起，他就真正能够形成企业的一个决策智能。比如说这个是类似的，就是这个我们当时回测过，还是比较有趣的。就是通过一家企业经营相关的这些专业数据做了整合之后，我们发现他有中标动作，他有专利动作，他有招聘动作，说明他的订单、研发和人才都在同时扩张。然后再结合这家公司过往几年的财务，发现这家公司它的收入情况也是在复合有增长的。那么这里面其实有一个判断，大概率这家公司未来还要再投两个亿，在产品研发，还有设备升级以及它产能建设这些维度。第三个就是它，这家公司大概率在三年内，它的产值是能够突破 10 个亿的。
诶，这个信息其实对于另一家，另一些企业来说，它也是有帮助的。对于那些做人才招聘的公司来说，诶，这家企业肯定有用工需求。对于做厂房扩建的中介来说，诶，这家企业有产能扩建的需求。甚至对于政府来说，这家企业有没有招商的需求：他原来在上海，我能不能把他招引到苏州去？把他招引到合肥去。
那每个地方其实都需要有商机和都需要有比较优质的企业，而 AI 呢，就能够通过这些专业数据去判断这家企业是不是一个优质的企业，以及是不是我需要的企业。
实际上这个我们测完之后发现，就是我们当时这批数据是从拿了他 2025 年 1 月份到 2025 年 6 月份，差不多应该有八九百条数据吧。测完之后，其实我们预测出来这家公司在 25 年的 6 月份之后，他一定会有厂房扩建的需求。但实际上后面我们发现他在公众号里面确实发了一个信息，说 2025 年的 6 月份，这个公司在上海松江一个地方就拿了一块新的地，就开工了。这个背后其实你会发现他企业的一些动作背后是有一些数据提供这个支撑的，所以呢就是 AI，当他有了智能，当他有了专业数据之后，他就能够把很多的事情看得见，到看得懂，并且到看得透，去进行一个迁移。
那其实很多的这个数据中呢，我觉得都是蕴藏很多企值得企业去关注的信号。比如说这个是我们给一家企业提供的一个商机挖掘的一个模块。我们帮助他去看清楚这家企业对您到底在干嘛？这家企业有什么样的痛点？以及这家企业到底关心什么？
这些所有的信号呢，其实都来自于什么？都来自于刚才我提到的专利、招聘，核心还是招聘，因为我发现企业很多它的增长需求、业务需求，其实都会反映到它的招聘信息里面。当我们把这些招聘的信息和数据都汇总完了之后呢，就能够形成这些多元的标签给到这个企业。这些信号的消费呢，其实它在很多其他的地方几乎都是找不到的。这个就是我觉得比较有意思的地方吧。因为我之前在帆软都是做战略出身，和都是做洞察的。其实以前做洞察的，说实话，是少数人的特权，因为你真正能够把一件事情从看得见到看得懂到看得透，这个跃迁的人，其实几乎是很少的。所以呢，我在思考， AI 加洞察能不能去解决企业增长问题，以及能不能把少数人的直觉变成全员的模式呢？其实我觉得是可以的。
从我刚才讲的那几个例子里面呢，其实都是类似的，又比如说这个，这家公司我简单讲一下吧。他们还是非常有趣的，属于很典型的在中国做了 20 年的一家制造业公司，他们差不多一年做 27 个亿的收入。但是你会发现中国现在整个企业，它面临的一个情况是增长是承压的，它很难增长。它的第一增长曲线已经饱和了，但它想要往第二增长曲线去开拓新的业务和开拓新的方向，但是它完全不知道新的业务和新的方向到底应该怎么做，就两眼一摸黑。
即便大家能够坐在一个会议室里面去讨论新的增长方向，但是是没有一个非常强的老师和没有非常强的数据来支撑他们的决策的。那当时我是给他们怎么做的？其实逻辑很简单，一个就是他们的新方向要做什么？OK，他们新方向现在是定位出来要做电池管理系统，你手机里面，你的电脑里面，你的无人机，你的汽车，只要有电池的地方，它就有电池管理系统，包括遥控器里面，我估计也有电池管理系统。那么他，我能够告诉他几个事情，第一个我就基于一个产业链的数据，告诉他全国做电池管理系统的公司一共有多少家？一共有 1053 家，这个就是你的潜在同行。
第二个，用到电池管理系统的下游的赛道分别有哪些？无人机、车子、电动车、摩托车，还有。这个飞机，还有电脑，还有手表等等等等。OK，你现在定位到哪个方向？他们现在定位到的是电动摩托车赛道。那么如果做定位到电动摩托车这个赛道，那么你的潜在市场，潜在客户到底有多少？以你在这个潜在客户和潜在这个市场里面，哪些是你值得去关注的企业？以及这些企业他最近有没有用电子管理系统的这个需求，都是可以通过刚才说的那些招投标啊，还有招聘啊，甚至联系方式这些维度都给拿到的。拿到了这些之后，其实就能够帮助他们快速从看清一个方向，到看到一个客户，咱再到看到。一个联系方式，这些就串通，串完之后呢，其实对他来说就是 AI 的帮助，就是能够实实在在的转化为可见的线索，而这些线索在到他们内部就可以转变为商机，商机进一步就能够变为...这个还是比较比较有趣的。
那另外的话就是类似，就是我觉得这个是我当时，就是我给自己做的一个，结果后来我发现就很多企业要的一个东西，这是什么呢？就是在中国，你想要去认识一个人，说实话你只能去找中间一个人给你微信拉个群。我 5 月份的时候在美国去了一趟 LinkedIn，然后我当时问他们为什么 LinkedIn 在中国做不起来？就说了一句话，就是中国本质上是一个熟人社会。你要在美国，你要去联系一个你不认识的人，你可以在 LinkedIn 上面去给他发一个消息，你只要非常真诚，他大概率是乐意来搭理你，并且是给你做转介绍的。
那如果说我今天想要认识货太网 Kevin，我跟他如果那个素不相识，其实不好意思，我是很难去认识他的。那最好什么？中间是有一个人，他既认识 Kevin，又认识我，他来拉这个群，才能够让我们两个有这个交集。
那所以呢，我又做了一件什么事情？
就是我每天差不多会去清，全网 200 万条左右的舆情信息。然后从这些舆情信息里面就清洗出来，企业 A 的张三跟企业 B 的李四，还有企业 C 的王五之间有什么样的关联关系？比如说我们今天办了这个活动，我是最后一个演讲的，刚才那位演讲嘉宾是第一个演讲的，有人第二个演讲，三个演讲，配套今天可能有个晚宴，这个活动呢，他大概率会在微信上去发一个推文，对吧？或者在视频号里面是有一些剪辑的。
我的这个 AI 呢，就能够帮助你去把这些非结构化的信息里面去提取出来，哪些企业有这个关联关系的？那么这些关联关系有了之后呢，其实你就可以问他，哎，我想认识企业 A 的张三,我同步认识这 100 个人是我的微信好友，能不能帮我去看看我的这 100 个微信好友能够进一步认识到哪些？我想认识的人，他就能够帮助你去实现这个人脉的穿透。这件事情其实在国内的很多的商业环境下，就能够帮助他们走得通。很多的工具只给你一个手机号之类的，你去打电话，说实话你肯定打不通。但是如果告诉你这些关联关系和企业的商务关系，你通过这些商务关系，你就能够大面积的，或者说非常高效的去完成一个陌生好友和陌生线索和商机的触达。这个其实也是我们当时因为有了这些专业数据之后，只能够做的一个事。
还有很多啊，就是包括我们去做一些简单的分析和竞对分析，还有行业分析的时候，就是其他的 AI 它就只能够通过公开的一些研报去走，但是我们去分析一家竞争对手，分析一个公司的时候，能够从它的专利库，比如这个，就直接去取它专利库的 377 条数据和它的招投标库里面的 2292 条数据，基于这个可信数据源去给你做的一家企业深度分析和行业深度分析的时候，它是绝对不会错的。
可能 AI 会错，但是基于这些可信数据做出来的报告和分析的话，理论上来说没有比这个更准和没有比这个更对的了。就比如说这个，这家公司呢，它是做二轮电动车的，他当时想要去研究自己的产品，用户是怎么评论的？用户。对于新产品和对于品类有什么样的关注点，所以当时直接就帮他取，就是他关注的一些型号和信息在抖音、微信、小红书里面 442 条，帖子，基于这 442 条帖子来对他的新品去看到了一些用户的信息和用户的痛点。因为他是基于 442 条可追溯的数据源所得出的分析和结论，这些结论其实也是站得住脚的。
又比如说是上海一家做接线端子的公司，他们其实竞争还是比较激烈的，就当时我们给他做的其实就是我给你全网，基于抖音专利、商标、招投标里面去看。就是你的同行有没有哪些新的产品出现？就比如说这个，当时有一个 MST，这个 2.5- 15 的这个型号，在 10 月 4 号的时候，在抖音里面就出现了 14 条相关的视频，并且这些视频曝光量差不多都在 5 万，这个说明什么？说明你的竞争对手在推这个新产品，并且这个新产品已经在开始铺货了。
所以呢，当底层有这些可行数据，并且在上层运用这个 AI 去做这些基于专业数据的监控之后，就能够帮助他们去实现一个比较好用的一个新品类，能够帮助他们更快的去看清市场的变化和看清竞争格局的变化。
类似的很多啊，今天因为时间关系，我不能过多的展开，我就用这一页来作为我今天分享的一个收尾吧。
其实我们发现，当一个 AI 底层具备了高质量的可行数据源，并且它又具备了专家级的分析经验和知识引擎，并且它有一个最前沿的智能，并且的话它又能够站在你自己的视角去做的时候，就能够实现从洞察入手，能够让这个洞察从少数的变成全员的模式。而这个全员的模式其实它是可以帮你去找到对的方向，找到对的客户，以及找到对的联系路径，并且把这些效率和你 token 的。支出正儿八经转化为线索，这些线索呢又能正儿八经转化为订单的，当它转变为订单之后，你的效率就能够变成利润。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
At the 1024 MCAI Day Master Forum, the co-founder of Shanghai Whale Slaughter Artificial Intelligence Ltd. and Vice-President of the Shang Soft Software Strategy, Shen Tao, shared the theme " AI Helping Enterprises Looking for Growth, Business Clients, Transforming Efficiency to Profit ".
Shen-tao asked a question enough to impress most business owners: "Ai, how much profit can it bring when employees do 1,000 more slides a year?"
In response, a chemical industry exchange was shared: two hours of R & D, production efficiency was discussed, while rest periods revealed that the chosen technology route may have lost competitiveness, which has increased his focus on one issue:
The wrong direction, AI might just let the business go the wrong way faster.
Shen-tao explained that their teams would use AI to provide a comprehensive analysis of recruitment, patents, bids and financial data from enterprises to identify potential business opportunities. For example, when an enterprise tendered for the purchase of 3150 kVA transformers, while a large number of test and test personnel were recruited, combined with relevant patent information, it could be assumed that it was expanding its capacity and that there was a need for enhanced quality control. This is a customer trail that deserves further follow-up for the equipment detector, data analyser and Manpower Services.
When a manufacturing enterprise explores BMS operations, the team first combs the industrial chain, peers and downstream markets, then screens potential customers and their immediate needs, and gradually puts the search for new growth points on specific clients. He also described AI ' s ability to extract connections between individuals and businesses from information such as activity reports and to help identify possible business referral routes.
In Shentao ' s view, a combination of professional data, business knowledge and AI analysis can provide more targeted support to enterprises in their choice of direction, client identification and business follow-up opportunities. It is also important to see if these leads can be translated into orders and profit.
The following is a complete summary of the Shentao lecture, which was revised and edited without changing its original intent.
Why don't they make more money?
First I'm gonna introduce myself, and I've been sailing softly for 10 years with data. In the past 10 years, we have found it difficult to really get into business to solve growth problems and solve real business problems. At best, it is an efficiency tool to help him solve this efficiency problem.
And this year I started starting a business out of soft sails, and I realized that AI came, that it could actually be a digital employee, or even an expert employee, to really help solve the business’s problems of profitability and growth.
And this year, I've actually been looking at a lot of companies, especially since I started studying AI from 23, but, honestly, it's been since this year that I've been in business since the beginning. And then, at the beginning of the year, I actually observed a very big phenomenon, which is whether it's working right now or not, or whether it's Doubao, and these personal-level AI tools are really hot, and, to be honest, it's a very simple thing for a company to do its own AI.
But when I'm talking to a lot of people, the number 1 and the boss, he's actually concerned about a problem, and it looks like I'm working with WorkBuddy, and it's getting shorter and shorter for my staff, but what's the point of making 1,000 slides for a year? Right? He can't solve this profit problem.
Including this March, when I was talking about a business, he was about 40 billion in revenue, 8000 people. After the fire of OpenClaw in February and March this year, a lot of OpenClaw was laid inside the company, spending almost 25% of their AI budget in a month. But after the AI budget was over, they built over 60 AI agents. More than 60 AI agents, which seem to be happy, actually the last count found that, uh, the final token input turned into time for employees to fish.
So, actually, I find that the personal effects are not necessarily equal to those of the organization, and even if you do, it is not necessarily equal to an increase in profits.
The whole business AI landed, honestly it was a very big system project.
For example, this company, which I used to know better, I won't say my name, but everyone knows it. You see, in the last five years, they've been doing very, very well, whether it's digital infrastructure or digital transformation, or AI's landing. They're all about a digitized infrastructure and digital capability, and there's AI, and they're building a system around explosives growth.
The system supports them 23 to 25 years of profit, and 23 to 25 years of income has doubled, and this is a good organization and business as compared to their peers. But if you look at his profits, you'll find his earnings are going up, but his profits are going down every year. So, after I've observed these phenomena, I've actually found that that's what I just mentioned. It's in a company where you're going to finish the AI, where you're going to be on the side of an employee, where you're going to improve personal efficiency, between individual efficiency and organizational efficiency, and after organizational efficiency, you're going to really make it a profit.
So, actually, I think in this year and three years, there's an AI down in the business, and there's only one major narrative logic, that's how you use AI to improve efficiency and turn it into profit. This mission and this question are actually my starting point. And I'm thinking about what kind of approach, and what kind of AI capacity that can help businesses turn this efficiency into real profit.
And before you answer that question, there's a very moving company that I've been visiting, and this is the company of Zhejiang in Li Shui, which earns almost six billion dollars a year and is very powerful in the chemical industry. We were at seminars like this, and then we talked about how AI could help me do research and development and production. Their background is that they saw the growth of the semiconductor and the new energy industry, and they saw the opportunity for the growth of new fluorine-containing materials for a new type of material, and made a strategic decision that I was going to develop a new product with fluorine.
So we have this seminar, and I'm like, "Ai, can you help me read my papers, help me decorate some chemical logic?" AI, can you help me improve the efficiency of the tests? AI, can you help me improve productivity? AI, can you help me with the equipment inspections? It was about two hours, two hours later, and it was a lot of talk about angent, but when we were talking about two hours of mid-term rest, you know that when a large company really talks about the core, it's probably not in the conference room, right?
So we were talking at the meeting, when the mid-time break was outside, and I actually asked one question. And that's where you've got the chance to look at new fluorinated materials, and we're going to discuss some of the things that AI is doing. But is the direction of new fluorinated products the right direction of opportunity? In fact, on the middle of the break, it was discovered that it had been eliminated by the market, by its peers and by its clients. So, it doesn't make any sense that they're working on a technology that's out of control.
So that's what I've been thinking about, and if you're in the wrong direction, you're going to talk about a lot of AI good things, and AI is going to make you worse.
So between the efficiency of the enterprise and the profits of the enterprise, there's a huge proposition that you have to do the right thing in the right direction, so you can turn the AI efficiency into a profit, otherwise AI will only lose more. Because there's a formula in it. It's a business profit. What is it? It's the right way to make the right decisions at the right time, and you're in the right direction, and you're going to do it in the right way to make your internal implementation more efficient. You can make your profit, you can make your efficiency that profit. So I'm actually gonna find out that now that AI's here, a lot of AI manufacturers and AI's providers, he's gonna be offering, that's a lot of digital employees. But I think that the digital employees are actually on the right half of this formula, which is to make your implementation more efficient. Is there an AI tool that can help you find the right time? Can I help you find the right direction? And I've had the experience of dealing with data for the last 10 years, and I've actually found that there's a lot of things in the data that are worth studying and studying.
So, I'm thinking about a problem right now, that AI is strong, and there's a lot of outside brains that can help you read materials, but it can read materials, and it's all public, and it can read things on Baidu, and it can read things that you can search for online, but it has some data sources that business really cares about, but AI is actually harder to read right now. For example, basic business information, for example, the financial data on listed companies in Wind, for example, data on professional bids, for example, data on direct recruitment by boss, for example, data on corporate risk-related claims, or even, because it was a company in Changzhou in the previous period, it also had some sort of affair, right? So like this, it's on REDnote, it's on Douyin, it's on WeChat, it's on closed ecology. You ask in Doubao that you might be able to ask about Douyin, you ask in Yuanbao that you might be able to ask about some data about WeChat, but you don't have that data on a platform that can shut down the ecology. And instead, I think that a business is going to judge a direction, to find the right customer, to find the right direction, and that's what is really needed.
So, I'm actually coming out right now, and I'm bringing together this vast variety of isomers that I think are really of professional value to businesses, and that you might need to open a VIP in a N place before, and I'm putting it all together. And when it comes to integration, there's one thing that's interesting, for example, if we buy this database, that's the kind of database that we're hiring, and there's a lot of interesting things going on in this database, like this company, this JD, this JD, this JD, this JD, what's his job? It is the responsibility of bringing subsidiaries to the market and integrating financial planning. You look at this data, and if you put it together by recruiting it instead of by reptiles, you'll find, uh, this company is going to be on the market. But you'll find a signal on the market that actually helps another company, right? So this is the one where, when you put all this data together, a lot of dimensions are visible to AI.
The second is AI, because it has a very strong intelligence, and it can understand some data sources. For example, if we have professional recruitment data, professional patent data, professional bidding data, you see this as actually interesting. You see this company in June 26th, '26, he had a factory, high voltage, full load, so he added a tender for a 3150 kVA transformer, which is a signal. And the other thing is that you'll find out that he's been recruiting inspectors, testers and co-inspectors in the last three weeks. You see these two data if they can really be digested by AI, actually AI can make some judgment.
What is this judgment?
First, hey, the factory is not enough, so he added a transformer, what does that mean? The company's production is expanding.
Secondly, it is in the process of expanding its capacity because he is recruiting inspectors and what does he say behind his back? It shows that he's under pressure. So, when AI went to read it, he could see that the company was expanding, and in the process, it was a combination of light modules, which he would apply for patents to protect.
Third, in the process of expansion, this capacity bottleneck has changed from capacity to good quality, so it is on a massive scale. So when this signal is actually read by AI, who will you find it useful? For those who do real estate, he's useful because he's going to expand, and I can give him services.
It would also be useful for a company that did data analysis, such as sails, because his production was expanding and he needed some quality control, board demand and data analysis. So, for those companies that do human services, well, he also needs it, because he has to work. Right? I can then provide other workers with services, as well as other testing equipment providers, which he needs.
So, after these signals actually turned into some of the signals and conclusions that AI could understand, he actually became a business opportunity for another company. So, what I found was that data combined with intelligence, and he could really be a decision-making intelligence for the firm. It's kind of similar, for example, and it's kind of interesting that we've looked back at it. It was through the consolidation of these professional data related to the operation of a business that we discovered that he had the winning moves, that he had the patent moves, that he had the recruitment moves, that his orders, research and development, and talent were expanding simultaneously. It then combines the company ' s financial performance over the past few years and finds that the company ' s revenue position is also growing at a compound. So there's a judgment that probably the company is going to invest another $200 million in future, in product development, in equipment upgrades and in capacity building. And the third is it, and it's probably in three years, and it's worth a billion dollars.
Well, it's actually helpful for another business, for others. For companies that recruit talent, this business must be useful. In the case of middlemen for plant expansion, there is a need for capacity expansion. Even for the government, is there a need for the company to recruit: could I recruit him to Suzhou when he was in Shanghai? Bring him to Fat Chai.
Every place needs business opportunities and higher-quality businesses, and AI can use these professional data to judge whether it's a high-quality business or not.
In fact, when we did this, it was when we took him from January 2025 to June 2025, about 800, 900. After the survey, we actually predicted that after 25 years in June, he would have needed to expand the plant. But in fact, we found out that he did send a message in WeChat Official Account, saying that in June 2025, this company took a new piece of land and started work in Shanghai Zhong River. And behind this is the fact that there's some data behind some of his company's moves, so, AI, when he's smart, when he's got professional data, he's able to see a lot of things, to understand them, to see them, to move them.
And that's a lot of data, and I think it's a lot of signals that companies need to look at. This, for example, is a module of a business opportunity that we offer to an enterprise. We helped him see what the company was doing to you. What kind of pain does this business have? And what does this company care about?
What are all these signals from? It comes from the patents, the jobs, the cores and the jobs that I just mentioned, because I find that many of the firms have their growth needs, their business needs, which are actually reflected in their recruitment information. When we're done with all this recruitment information and data, it is possible to create these multiple labels for the enterprise. And the consumption of these signals is almost impossible to find in many other places. This is where I find it interesting. Because I used to be strategic and perceptive. In fact, it was a privilege of a few who used to be perceptive, to be honest, because you were truly able to see one thing from sight to light, and the person who jumped almost rarely. So I'm thinking, AI, if you can figure out if you can solve the growth problem of a business, and if you can turn the instincts of a few into a full-fledged model? Actually I think it's okay.
From the examples I've just given, it's all similar, for example, this company, let me just say. They're also very interesting, and they're typical manufacturing companies in China for 20 years, and they do about 2.7 billion dollars a year. But you'll find that the whole business in China is now facing a situation where growth is under pressure and it's hard to grow. Its first growth curve was saturated, but it wanted to open up new business and new directions to the second growth curve, but it had no idea what the new business and new direction should be, and it had black eyes.
Even if you can sit in a conference room to discuss new directions for growth, there is no very strong teacher and no very strong data to support their decision-making. What did I do to them? It's a simple logic. What's their new direction? OK, their new direction is now to locate the battery management system, inside your phone, inside your computer, in your drone, in your car, where there's a battery, there's a battery management system, including a remote control system, and I think there's a battery management system. So he, I can tell him a few things, and the first I can tell him, based on data from an industrial chain, how many companies do battery management nationwide? One in 1053, this is your potential companion.
Second, what are the different tracks downstream of the battery management system? Drones, cars, electric cars, motorcycles, and... The plane, the computer, the watch, etc. Okay, where are you headed now? They're now on the motorbike track. So if you're on the track of an electric motorcycle, how many potential customers do you have? With you in this potential customer and this potential market, which are your businesses worthy of attention? And the need for these companies to use electronic management systems in the recent past can be obtained through the bidding and recruitment and even the contact details. When they get these, they can actually help them quickly to see one direction, to see a client, to see one. One way of connecting, and then after that, is to him that AI can actually turn into visible clues that can be turned into business opportunities within them, and that can be more interesting...
And the other thing is similar, which is that I think it was one of the things I made myself at the time, and then I found out that there was only one thing that many companies wanted. Just in China, you want to meet someone, to be honest, you can only find someone in the middle to pull a bunch of WeChats. I went to LinkedIn in May, and I asked them why LinkedIn couldn't do it in China? Just one thing: China is essentially a familiar society. You're going to America, you're going to contact someone you don't know, you're going to send him a message on LinkedIn, and if you're being very sincere, he's probably willing to come and talk to you and give you an introduction.
So if I wanted to meet Kevin today, if I don't know him, I'm sorry, but I can't. What's the best? There was a man in the middle who knew Kevin and me, and he pulled this group so that we could have this intersection.
So, what did I do?
It's just that I'm almost clear every day, about 2 million love messages. So, what's the connection between Company A and Company B and Company C? For example, we had this event today, and I was the last to speak, and the speaker was the first to speak, and there was a second speech, three speeches, and a dinner that might accompany us today. Or there's some clippings in the video number.
This AI of mine can help you to extract this unstructured information, which companies have this connection? So when these connections are made, you can actually ask him, hey, I want to know Zhang San of Company A, I want to know this 100 people in sync. I want to know someone, and he can help you to get through that person's veins. In fact, in many domestic business environments, this can help them get through. Lots of tools just give you a cell phone number or something. You go on the phone and you'll never get through. But if you are told about these connections and business relationships, through them, you will be able to do a lot, or very efficiently, of a strange friend and a strange clue and business opportunity. This is actually the only thing that we could have done because we had these professional data.
There are a lot of other things, including simple and competitive analyses, and industry analyses, and other AIs, which can go only through open research, but when we analyze a competitor, analyze a company, we can go directly from its patent pool, for example, 377 data from its patent bank and 2292 data from its bid bank, based on this credible source of data, and it's absolutely not wrong when one of the firms and industry depth analyses are done.
Perhaps AI will be wrong, but there is no better and no better report and analysis based on these credible data than this. What about this company, for example, he's a two-wheeled electric car, and he was trying to study his product. User. The attention given to new products and products was such that he was given direct access to some of his models and messages in Douyin, WeChat, REDnote, and the post on the basis of which 442 posts were sent to see some user information and user pains. Because of his analysis and conclusions based on 442 traceable sources of data, they are in fact valid.
For example, a Shanghai-based company that actually competed hard, and what we did to him was I gave you the Internet based on Douyin patents, trademarks, bids. Is there anything new coming out of your people? So, for example, there was this MST, this 2.5-15 model, and on the 4th of October, there were 14 related videos in Douyin, and they were almost 50,000. That means your competition is pushing this new product, and this new product is already starting to work.
So, when these viable data are available at the bottom, and when AI is used at the top level to perform these professional data-based monitoring, it can help them to achieve a new and better-used product that can help them to look faster at changes in the market and at changes in competition patterns.
A lot like this, I cannot go on too much today because of time, so I will use this page as a wrap-up for what I have shared today.
In fact, we find that when an AI bottom has a high-quality and viable source of data, and it has an expert-level analytical experience and knowledge engine, and it has a cutting-edge intelligence, and it can do it from your own perspective, it can achieve a perceptive model that allows insight from a few to full capacity. And this whole staff model actually helps you find the right direction, find the right client, find the right connection path, and put these efficiencies and you token. The expenditure is translated into clues, and these clues are transformed into orders, and when it is converted into orders, your efficiency becomes profit.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.