[{"data":1,"prerenderedAt":22},["ShallowReactive",2],{"article-small-business-internal-tools-with-ai":3},{"slug":4,"date":5,"read_time":6,"image":7,"cta":8,"zh":10,"en":16},"small-business-internal-tools-with-ai","2026-09-02",8,"/logo.svg",{"link":9},"https://ai.xfeixie.com/",{"title":11,"tag":12,"excerpt":13,"keywords":14,"content":15},"中小企业该自己做内部工具，还是继续用通用软件","行业实践","通用软件让你迁就它的流程，定制开发太贵。当做一个内部工具的成本降到一天，这道选择题的答案变了。","企业内部工具,零代码开发,AI生成APP,审批系统,中小企业数字化","\n\u003Cp>几乎每家几十人的公司都经历过同样的循环：用 Excel 太乱，买通用软件用不顺，找外包报价二十万，最后回到 Excel 加微信群。\u003C/p>\n\u003Cp>这个循环的成因很简单：过去做一个贴合自己流程的内部工具，成本远高于忍受不合适的方案。当这个成本降到「一个人一天」，选择就变了。\u003C/p>\n\u003Ch2>通用软件为什么用不顺\u003C/h2>\n\u003Cp>不是软件做得差，是它必须服务一万家公司。所以它的请假流程是所有公司请假流程的平均值，你的公司恰好有个特殊环节——比如车间主任要先确认排班——它就装不下。\u003C/p>\n\u003Cp>结果是：一半流程在系统里，一半在群里。数据两头都不全，管理层看到的报表也就不能当真。\u003C/p>\n\u003Ch2>自己做的三个真实好处\u003C/h2>\n\u003Ch3>1. 流程按你的实际情况走\u003C/h3>\n\u003Cp>「请假先给车间主任确认排班，再给厂长审批，超过三天同步给 HR」——这就是一句话的需求。不用为了迁就系统去改公司的做事方式。\u003C/p>\n\u003Ch3>2. 字段是你要统计的字段\u003C/h3>\n\u003Cp>通用软件的报表给你它认为重要的指标。自己做的工具，字段就是你月底真正要看的那几个数，导出即可用，不用再手工二次加工。\u003C/p>\n\u003Ch3>3. 改起来不用等版本\u003C/h3>\n\u003Cp>流程调整了，说一句话改掉。不用提工单、不用等厂商下个季度的迭代。\u003C/p>\n\u003Ch2>哪些内部工具最值得先做\u003C/h2>\n\u003Cp>按投入产出排，通常是这个顺序：\u003C/p>\n\u003Cul>\n\u003Cli>\u003Cstrong>正在用纸质或 Excel 手工汇总的\u003C/strong>：巡检表、报工单、盘点表。改成手机填报，省下的是每月固定的人工工时。\u003C/li>\n\u003Cli>\u003Cstrong>正在靠微信群传递的审批\u003C/strong>：请假、报销、用章申请。上系统之后至少有记录可查。\u003C/li>\n\u003Cli>\u003Cstrong>需要给外部客户填的\u003C/strong>：随访问卷、预约登记。发链接就能填，不用让客户装东西。\u003C/li>\n\u003C/ul>\n\u003Cp>反过来，涉及资金结算、税务合规的核心系统，不建议一上来就自己做——这类场景对准确性的要求应该由专业系统和专业人员承担。\u003C/p>\n\u003Ch2>落地时的三条建议\u003C/h2>\n\u003Ch3>先做一个环节，不要做一个平台\u003C/h3>\n\u003Cp>想一次做出「公司统一管理平台」的，基本都做不完。先把巡检表这一个环节做完、让一线用起来，比画一张大图有用得多。\u003C/p>\n\u003Ch3>让真正干活的人来提修改\u003C/h3>\n\u003Cp>需求方自己就能改，是这套做法最大的价值。让车间主任自己说「这个字段用不上，删了」，比经过三层转述准确得多。\u003C/p>\n\u003Ch3>先跑两周再谈推广\u003C/h3>\n\u003Cp>一个环节、五个人、两周。跑通了再往其他部门复制，此时你已经知道哪些设计是想当然的。\u003C/p>\n\u003Ch2>成本对比\u003C/h2>\n\u003Cp>过去：外包定制 10-30 万，周期 2-3 个月，改需求另算钱。\u003Cbr>现在：一个懂业务的人，一天做出第一版，两周内迭代到能全员使用，年费不到千元。\u003C/p>\n\u003Cp>差别不只是省钱，是让「不值得为它专门立项」的那些小工具终于有机会被做出来——而公司里的效率损耗，恰恰大多藏在这些小地方。\u003C/p>\n",{"title":17,"tag":18,"excerpt":19,"keywords":20,"content":21},"Should a Small Company Build Its Own Internal Tools or Keep Buying Generic Software?","Practice","Generic software makes you bend to its process; custom development costs too much. When building an internal tool drops to a day of work, the answer changes.","internal tools, no code development, AI generate app, approval workflow, small business software","\n\u003Cp>Almost every company of a few dozen people runs the same loop: spreadsheets get messy, generic software does not fit, an agency quotes a fortune, and everyone falls back to spreadsheets plus a group chat.\u003C/p>\n\u003Cp>The cause is simple. Building an internal tool that matched your process used to cost far more than tolerating one that did not. When that cost drops to one person for one day, the trade-off flips.\u003C/p>\n\u003Ch2>Why generic software does not fit\u003C/h2>\n\u003Cp>It is not badly built - it has to serve ten thousand companies. Its leave-approval flow is the average of all their flows, so your one unusual step (say, the shop supervisor confirming the shift roster first) has nowhere to live.\u003C/p>\n\u003Cp>The result: half the process is in the system and half is in a chat group. Neither side has complete data, so management reports cannot be trusted.\u003C/p>\n\u003Ch2>Three real advantages of building it yourself\u003C/h2>\n\u003Ch3>1. The process follows your reality\u003C/h3>\n\u003Cp>\"Leave requests go to the shop supervisor to confirm the roster, then to the plant manager, and anything over three days is copied to HR\" is a one-sentence requirement. You stop reshaping the company to fit the software.\u003C/p>\n\u003Ch3>2. The fields are the ones you report on\u003C/h3>\n\u003Cp>Generic software reports the metrics it thinks matter. A tool you built captures the handful of numbers you actually review at month end - exportable as-is, with no manual rework.\u003C/p>\n\u003Ch3>3. Changes do not wait for a release\u003C/h3>\n\u003Cp>The process changed? Say one sentence and it is changed. No support ticket, no waiting for a vendor's next quarterly release.\u003C/p>\n\u003Ch2>Which internal tools to build first\u003C/h2>\n\u003Cp>Ranked by payoff, usually:\u003C/p>\n\u003Cul>\n\u003Cli>\u003Cstrong>Anything still totalled by hand on paper or in spreadsheets\u003C/strong>: inspection sheets, shop-floor reports, stock counts. Moving them to mobile forms saves a fixed number of hours every month.\u003C/li>\n\u003Cli>\u003Cstrong>Approvals currently passing through group chat\u003C/strong>: leave, expenses, stamp requests. Even a basic system gives you a searchable record.\u003C/li>\n\u003Cli>\u003Cstrong>Forms external customers fill in\u003C/strong>: follow-up surveys, appointment intake. Send a link - nothing for the customer to install.\u003C/li>\n\u003C/ul>\n\u003Cp>Conversely, do not start with core systems for financial settlement or tax compliance. Those accuracy requirements belong with specialist systems and specialist people.\u003C/p>\n\u003Ch2>Three pieces of advice\u003C/h2>\n\u003Ch3>Build one step, not a platform\u003C/h3>\n\u003Cp>Anyone trying to build \"the unified company platform\" in one go does not finish. Completing the inspection sheet and getting the frontline to use it beats drawing a grand diagram.\u003C/p>\n\u003Ch3>Let the people doing the work request changes\u003C/h3>\n\u003Cp>The biggest gain here is that the person with the requirement can make the change. The shop supervisor saying \"we never use this field, drop it\" is far more accurate than the same request relayed through three layers.\u003C/p>\n\u003Ch3>Run it for two weeks before rolling it out\u003C/h3>\n\u003Cp>One step, five people, two weeks. Copy it to other departments after that - by then you know which of your assumptions were wrong.\u003C/p>\n\u003Ch2>The cost comparison\u003C/h2>\n\u003Cp>Before: a custom build from an agency, two to three months, and changes billed separately.\u003Cbr>Now: one person who understands the business ships a first version in a day, iterates to company-wide use within two weeks, for a few hundred dollars a year.\u003C/p>\n\u003Cp>The difference is not only money. It is that the small tools never worth a formal project finally get built - and most of a company's wasted effort hides in exactly those small places.\u003C/p>\n",1791342084608]