Netic：一张高额电费单催生的 4.5 亿美元 AI 收入引擎
Netic: A $450M AI Revenue Engine Born From a High Electricity Bill
一张电费单，捅出一个 5000 亿美元的漏洞
Melisa Tokmak 毕业于斯坦福，先后在 Facebook 和 Scale AI 做产品，是 Scale AI 政府与企业业务的搭建者。她和丈夫在加州买房后，收到一张几千美元的电费单。
她想弄清楚为什么这么贵——是暖气？隔热？还是线路问题？她给好几家公司打电话。多数不接。少数接了，记不住她的名字和地址。 好不容易安排了上门，她在家等了好几个小时，烧着更多的电费，不知道人来不来。最后发现是暖气的问题。
这张电费单，催生了 Netic。
Tokmak 后来发现，这不是她一个人的遭遇。美国的水管工、电工、暖通空调工人、屋顶工——这群人支撑着一个约 5000 亿美元的市场，但他们的生意极度依赖电话。一个水管工一天电话响 47 次，能接 12 次，剩下 35 次就是丢掉的客户、丢掉的收入。 师傅不能一边到房子底下修水管，一边接前台电话。
这个行业的痛点极其原始：不是技术问题，是没人接电话。
而硅谷看不上这块市场。过去几年 AI 圈卷的是「给白领做 Copilot」，没人愿意碰「让 AI 接水管工电话」这种脏活。
Netic 的关键定义：AI 收入引擎，不是 AI 前台
2026 年 7 月，Netic 登上 Forbes Next Billion-Dollar Startups 榜单：累计融资 4300 万美元，估值 4.5 亿美元，领投方是 Founders Fund 和 Greylock。B 轮完成时，全公司只有 22 个人。
很多人以为 Netic 就是一个 AI 前台，帮小企业接电话。这么说没错，但只说对了一半。
Netic 把自己定义为「AI 收入引擎」，不是「AI 前台」。 这两个词的区别，是理解这家公司的关键。
它有两个核心产品：
Convert：把每一个电话变成收入
Convert 负责转化。AI 代理跨电话、短信、网页聊天所有渠道处理客户需求。它不只是回答问题，而是完成整个工作流——资格认定、报价、排程、写入系统，全程不需要人插手。
Netic 官方的说法是「100% 处理所有渠道的需求」。
一个典型场景：2025 年圣路易斯遭遇龙卷风袭击，一家客户的屋顶维修业务在 90 分钟内涌进 600 多个电话。Netic 的 AI 自主接了绝大部分，做紧急程度分流和预约排程。 一家 HVAC 企业在旺季把溢出和夜间电话切给 Netic 后，避免了呼叫中心人员翻倍，保持了 90% 以上的预约转化率，客单价提升了 1.6 倍。
Cultivate：在客户自己还不知道之前，主动出击
Cultivate 负责培育，是一个主动出击的工具。
它分析客户数据和外部信号——比如天气、季节、设备寿命——在客户自己意识到需要之前就主动联系。比如识别出一批设备老化的客户，在寒潮来临前定向触达，帮他们填满淡季的预约表。
这不是发营销邮件，而是：
- 识别信号（设备年龄、上次维修时间、当地天气预报）
- 生成个性化外呼话术
- 主动拨出，完成排程
- 结果写入 CRM
Cultivate 的逻辑是：不要等客户出问题才来找你，在问题发生前就锁定这笔订单。
技术核心：Netic Brain
两个产品的底层是一个叫 Netic Brain 的模型，专门针对服务业工作流训练。
它需要处理：
- 浓重的地区口音（南方口音、移民口音等）
- 情绪激动的来电者（设备坏了、房子漏水的紧急状态）
- 复杂的排程逻辑（工人技能匹配、行程优化、材料库存）
官方工程博客的描述是：「一个紧急问题的来电者，应该感觉自己是在和真正懂行的人说话。」
这是通用大模型难以直接胜任的场景——不只是语言理解，还需要深度融合服务业的业务逻辑和领域知识。Netic Brain 是在这个垂直场景上的专项训练。
价值创造的完整链条
错过的电话 → 丢失的客户
↓ (Netic Convert 介入)
接住所有渠道来电 → 完成预约转化 → 收入捕获
休眠客户 + 外部信号
↓ (Netic Cultivate 介入)
主动外呼 → 预约锁定 → 客单价提升 + 淡季填充核心公式：减少漏单 × 提升转化 × 主动创造需求 = 收入增长
这不是「帮你接电话节省人力成本」，而是「帮你捕获原本根本不存在于你收入里的那 35 个电话」。
为什么 22 人能做到 4.5 亿估值
几个结构性原因：
1. 市场足够大且足够原始
5000 亿美元的美国服务业，接电话是最核心的业务入口，但数字化程度极低。这个空间里几乎没有真正的技术竞争对手——原来的「解法」就是多雇一个前台。
2. 定位是收入，不是成本
「帮你省前台工资」和「帮你多赚 35 个订单」，定价逻辑完全不同。后者可以按转化收入分成，ROI 更容易量化，销售阻力更小。
3. 专项模型创造护城河
Netic Brain 是在海量服务业通话数据上训练的。每接入一家新客户，就有更多语料；语料越多，模型越准。这是一个数据飞轮，通用大模型难以快速复制。
4. 创始人背景精准匹配
Scale AI 企业业务的搭建者——她做的就是「把 AI 能力卖给对 AI 一无所知的传统企业」。这个经历直接迁移到了 Netic 的销售路径。
可复制的「AI 接电话」收入转化模式
Netic 的模型可以迁移到其他依赖电话的服务业场景：
| 场景 | 痛点 | Convert 对应 | Cultivate 对应 |
|---|---|---|---|
| 汽车修理厂 | 技师忙时错过预约 | AI 接报修电话，完成工单 | 保养周期到期前主动提醒 |
| 宠物诊所 | 夜间/节假日无人接听 | 全天候接诊咨询+排程 | 疫苗到期、复查时间主动外呼 |
| 牙科诊所 | 爽约率高 | 自动确认+改期处理 | 定期清洁提醒，填充空档期 |
| 律师事务所 | 首次咨询转化低 | 资格初筛+预约律师 | 案件进展节点主动联系 |
| 家政服务 | 旺季接单混乱 | 全渠道接单+自动排程 | 节前主动推套餐，填满档期 |
| 健身房 | 新会员流失快 | 试课预约全自动化 | 活跃度下降时主动挽留 |
三个前提条件：
- 业务高度依赖电话/预约
- 有明确的「接到就能转化，没接到就丢单」的漏洞
- 客户有可预测的重复需求周期（设备维保、健康检查等）
满足这三点的行业，Netic 模式的本质都能复制：把每一个接触点变成收入节点，把可预测的需求周期变成主动创收机会。
总结
Netic 做的事情在表面上很简单——「让 AI 帮你接电话」——但它找到的是一个被硅谷忽略的 5000 亿美元市场，在里面做了一件本质上更重要的事：把接电话从成本中心变成收入引擎。
22 人，4.5 亿美元估值，Founders Fund + Greylock 押注，Forbes 榜单。它的下注是：在 AI 圈争相给白领做 Copilot 的时候，服务于真正的基础设施——那群每天爬屋顶、钻管道、修空调的人。
官网: usenetic.com
创始人: Melisa Tokmak（前 Scale AI 政府与企业业务负责人）
投资方: Founders Fund · Greylock
原文作者: @做战略的Ray（小红书）
Netic: A $450M AI Revenue Engine Born From a High Electricity Bill
Melisa Tokmak — Stanford graduate, product leader at Facebook and then Scale AI, where she helped build the government and enterprise business — moved into a California home with her husband and received a utility bill for thousands of dollars.
She wanted to know why. Heating? Insulation? Wiring? She called several companies. Most didn’t pick up. The ones that did couldn’t remember her name or address. After finally scheduling a visit, she waited at home for hours — running up more electricity — unsure if anyone was coming. It turned out to be a heating issue.
That electricity bill created Netic.
Tokmak soon realized this wasn’t just her problem. America’s plumbers, electricians, HVAC technicians, and roofers support roughly a $500 billion market — a market that runs almost entirely on phone calls. A plumber receives 47 calls a day and can answer 12. The other 35 are lost customers, lost revenue. A technician under a house fixing pipes can’t simultaneously work the front desk.
The industry’s problem is brutally simple: not a technology problem — a nobody-answers-the-phone problem.
And Silicon Valley wasn’t interested. The AI industry spent years racing to build Copilots for white-collar workers. Nobody wanted to touch “AI answering calls for plumbers.”
Netic’s Key Distinction: AI Revenue Engine, Not AI Receptionist
In July 2026, Netic appeared on the Forbes Next Billion-Dollar Startups list: $43M raised total, $450M valuation, led by Founders Fund and Greylock. At Series B close, the whole company was 22 people.
Most people assume Netic is an AI receptionist — it answers phones for small businesses. That’s not wrong, but it’s only half the story.
Netic defines itself as an “AI Revenue Engine,” not an “AI front desk.” That distinction is the key to understanding the company.
It has two core products.
Convert: Turn Every Call Into Revenue
Convert handles inbound conversion. AI agents handle customer demand across every channel — phone, SMS, web chat — simultaneously. They don’t just answer questions; they complete the entire workflow: qualification, quoting, scheduling, and system entry, without human intervention.
Netic describes this as “handling 100% of demand across all channels.”
A concrete example: when a tornado hit St. Louis in 2025, a roofing customer received 600+ calls in 90 minutes. Netic’s AI autonomously handled the vast majority — triaging urgency and booking appointments. An HVAC company that routed overflow and after-hours calls to Netic avoided doubling their call center staff during peak season, maintained 90%+ booking conversion, and increased average order value by 1.6×.
Cultivate: Reach Out Before the Customer Knows They Need You
Cultivate handles proactive growth. It analyzes customer data and external signals — weather forecasts, seasons, equipment age — and reaches out before the customer realizes they have a need.
For example: identify customers with aging equipment, reach them before the cold snap arrives, and fill the slow-season calendar with pre-booked appointments.
This isn’t marketing email blasts. It’s:
- Signal detection (equipment age, time since last service, local weather)
- Personalized outbound scripts
- Autonomous outbound calls that complete scheduling
- Results written back to CRM
Cultivate’s logic: don’t wait for the customer to have a problem — lock in the job before the problem arrives.
Technical Core: Netic Brain
Both products run on an underlying model called Netic Brain, trained specifically on service-industry workflows.
It needs to handle:
- Heavy regional accents (Southern US, immigrant communities)
- Distressed callers (flooded basement, broken AC in summer heat)
- Complex scheduling logic (technician skill matching, route optimization, parts inventory)
From Netic’s engineering blog: “A caller with an urgent problem should feel like they’re talking to someone who actually knows the job.”
This is where general-purpose LLMs fall short — it’s not just language understanding, it requires deep integration of service-industry business logic and domain knowledge. Netic Brain is purpose-built for this vertical.
The Full Value Creation Chain
Missed calls → Lost customers
↓ (Netic Convert)
Capture all inbound demand → Complete booking → Revenue captured
Dormant customers + external signals
↓ (Netic Cultivate)
Proactive outbound → Lock in appointments → Higher order value + filled slow seasonCore formula: Reduce missed calls × Improve conversion × Proactively create demand = Revenue growth
This isn’t “save money by replacing a receptionist.” It’s “capture the revenue from those 35 calls that never existed in your books.”
Why 22 People Can Hit $450M Valuation
Several structural reasons:
1. Large market, low digitization
A $500B US services market where phone calls are the primary business entry point — and digital infrastructure is minimal. The “solution” before Netic was simply hiring another receptionist.
2. Revenue positioning, not cost positioning
”Save receptionist salary” and “capture 35 more orders per day” are priced completely differently. The latter can charge as a percentage of converted revenue — ROI is quantifiable, sales friction is lower.
3. Vertical model as a moat
Netic Brain trains on service-industry call data. Every new customer adds more training signal; more signal makes the model sharper. A data flywheel that general-purpose models can’t easily replicate.
4. Founder background precisely matched
Scale AI enterprise business builder — her job was “sell AI capabilities to traditional businesses that know nothing about AI.” That skill transfers directly to Netic’s sales motion.
Replicating the “AI Phone Call” Revenue Model
Netic’s approach can be applied to any phone-dependent service business:
| Sector | Problem | Convert analog | Cultivate analog |
|---|---|---|---|
| Auto repair | Missed bookings during busy hours | AI handles repair intake and work orders | Proactive maintenance reminders before due date |
| Veterinary clinics | After-hours/holiday calls unanswered | 24/7 appointment intake | Vaccine reminders, follow-up scheduling |
| Dental offices | High no-show rates | Automatic confirmation and rebooking | Periodic cleaning reminders, slot filling |
| Law firms | Low first-consultation conversion | Initial screening + attorney scheduling | Case milestone outreach |
| Home services | Chaotic peak-season booking | Omni-channel intake + auto scheduling | Holiday package push, calendar filling |
| Gyms | Fast new member churn | Trial class booking automation | Engagement drop triggers proactive retention |
Three prerequisites for replication:
- Business fundamentally driven by phone calls / appointments
- Clear “answered = converted, missed = lost” leakage in the funnel
- Customers have predictable repeat need cycles (maintenance, health checks, seasonal service)
Any industry satisfying these three conditions can use Netic’s model: turn every contact into a revenue node, turn predictable demand cycles into proactive revenue creation opportunities.
Summary
What Netic does sounds simple on the surface — “AI answers your phone calls.” But it found a $500B market that Silicon Valley ignored, and did something far more important inside it: turned phone-answering from a cost center into a revenue engine.
22 people, $450M valuation, Founders Fund and Greylock, Forbes list. The bet: while every AI company races to build Copilots for knowledge workers, serve the actual infrastructure — the people climbing roofs, crawling through pipes, and fixing air conditioning every day.
Website: usenetic.com
Founder: Melisa Tokmak (former Scale AI government & enterprise lead)
Investors: Founders Fund · Greylock
Original analysis: @做战略的Ray (Xiaohongshu)
关于本站 · 免责声明
🍄 Mushroom Research Blog 是非营利、免费公开的个人科技观察博客与公众号 XStack18，不接受商业合作、不代表任何企业或机构立场，也不谋求商业利益。我们以个人视角客观中立地记录和分析 AI、Web3 等领域的最新模型发布与技术动态——不止转述新闻标题或二手信息，而是给出有独立思考的深入分析，希望帮更多人获得有价值的一手科技认知。
⚠️ 文中介绍的开源代码与模型，仅供学习交流与技术借鉴。它们大多仍处于早期阶段，有待进一步研究和验证，请勿直接用于工作或生产环境；如需采用，请先自行充分测试，并核实其许可证与安全性。
Open-source code and models featured here are shared for learning and reference only. Most are early-stage and still need further study and verification — please don't use them directly in your work or in production. Test them thoroughly and check their licenses and security first.
- 本站文章均为作者基于公开信息的个人研究与观点整理，不代表文中提及的任何公司、产品、模型的官方立场，未与其构成商业关联或合作关系。
- 科技行业信息更新极快，我们尽力保证内容准确、及时，但不对完整性、实时性做绝对保证，具体请以相关企业/项目官方公告为准。
- 文中引用的第三方商标、产品名称、图片、数据等版权归原权利人所有，我们会尽量注明来源；如你认为存在版权疑问或侵权，请通过下方邮箱联系我们，收到通知后会尽快核实处理（更正、加注来源或删除）。
- 文章内容仅为技术科普与个人观点，不构成投资、法律或其他专业建议，据此进行任何决策的后果需自行判断和承担。
📮 侵权 / 勘误 / 合作咨询：hello@mushroom.cv
💬 评论与讨论
使用 GitHub 账号登录后发表评论