Building an Autonomous Enterprise for Real-World Services with Netic Founder Melisa Tokmak

Summary

• Netic is building an “autonomous enterprise” for essential-service operators, automating everything around the physical service while leaving delivery and human labor to people. Its agents handle voice, text, and web interactions, infer customer needs, apply operational rules, and deploy the right worker. The ambition is AI that can “run millions of real-world businesses that keep the world running.”
• The wedge has moved well beyond call overflow: Tokmak says over 70% of customers are now “Netic first,” with AI handling every customer’s initial interaction. In HVAC, that means determining equipment type, urgency, serviceability, customer lifetime value, and whether a scarce boiler specialist should go today or later—not merely answering the phone.
• Tokmak chose scalable vertical software over an AI-enabled services roll-up because M&A is not her skill set and roll-up software is generally captive to the acquired assets. Her alternative is a common intelligence layer on which “every real-world business can run,” letting operators concentrate on service quality and differentiated labor.
• Tokmak’s differentiation thesis rests on vertical execution rather than models alone. She does not see the leading labs as a competitive risk: they pursue generalizable problems, while mission-critical services require domain focus plus “harnesses and orchestration, the software and the product” across three layers. Waiting for AGI to solve essential services is, in her words, “operationally and intellectually a bit lazy thinking.”
• The adoption evidence challenges the assumption that essential-services businesses are technology laggards. Tokmak cited one $500,000 contract completed end-to-end in 14 days and estimated that Netic has generated “over $600 million” for customers from AI-handled interactions. Her commercial pitch is primarily net-new revenue, because “it would be pretty sad if we used AI only for cost cutting.”
• The company’s operating philosophy favors durable craftsmanship over short-term AI opportunism. Tokmak avoids “shiny object seekers” and screens for agency demonstrated repeatedly over a lifetime, coupled with follow-through, rigor, and patience. She connects founders’ fear of lab roadmaps to businesses designed for quick exits rather than “dedicating your whole life” to a company for decades.
• Tokmak’s broader AI optimism centers on access to education, but she does not confuse access with action. Putting knowledge “in your pocket” removes resources as a constraint and makes the world more about agency; the darker caveat is that she guarantees most people still will not choose to act, however easy technology makes that choice.

Deep dive

1. Essential services require orchestration, not another chatbot

• Tokmak defines Netic as AI for large enterprises spanning HVAC, plumbing, electrical work, roofing, automotive, hospitality, pet care, and consumer wellness. “Netic exists between the company and its customers,” interpreting demand and matching it against each operator’s rules, serviceability, and labor.

• Before Netic, these workflows were handled primarily by people. Many operators are EBITDA businesses, often private-equity-owned, and cannot grow without continually investing in staff. Their support teams can be unreliable: a large business may start before 6 a.m., face several quits or no-shows, and then absorb a surge during a heat wave or other seasonal peak. Because many services are commodities, failing to answer can send a customer to the next search or AI result.

• Her load-bearing example starts at minus 20 degrees when someone’s heat fails. Netic can answer by voice, text, or web, retrieve household records, understand the equipment and urgency, and decide whether the company can help—especially when a child or elderly person may be inside.

• The allocation problem goes deeper than sending the next available technician: which units are installed, whether service is needed today or tomorrow, the customer’s lifetime value, and whether to deploy a scarce specialist who handles boilers or newer systems. The objective is both “customer delight” and more revenue.

• Elad’s initial framing was call overflow; Tokmak’s correction matters. Overflow is how many customers begin, but over 70% are now “Netic first,” or “N1,” meaning the first interaction with the company is with a Netic agent rather than a human team.

2. Software scale beat buying and operating the underlying assets

• Tokmak’s motivation combines personal history with a technical problem. Raised in a small Turkish town where people commonly work in these industries, she came to the United States for college on a full Stanford scholarship and had never owned a computer. After four years building government and enterprise businesses at Scale AI, she wanted technology that served the real world and “a product that can scale and compound.”

• Elad’s pushback—worth keeping—was why not buy HVAC and similar operators, then optimize them with AI. Tokmak’s answer had three parts: M&A is not her core skill, she is “a builder” and product engineer, and roll-up technology remains captive to the few acquired companies rather than becoming infrastructure for an entire market.

• Robotics occupies “the same book maybe, but different chapters.” Tokmak expects a future robotics chapter, but says many service trades remain far away: every building would need to be 3D-printed or completely standardized, while existing tasks demand dexterity across tiny spaces and varied parts. Technicians may need to open walls before even knowing the problem.

3. Vertical execution fills the gap left by general-purpose labs

• Asked whether OpenAI, Anthropic, Google, or Meta could build the same product, Tokmak chuckled at the recycled form of “can Google do this?” She regards the leading labs as strong businesses and partners, but not a competitive risk for Netic’s focused enterprise workflow.

• Her focus argument cuts both ways: OpenAI builds products quickly but also “kills them really fast,” while Anthropic’s celebrated coding focus contrasts with what she sees in the enterprise case—roughly 20 products rather than a similarly focused offering. Essential-service operators need a stable, long-term relationship, not simply rapid product proliferation.

• Researchers naturally seek the most generalizable solution, Tokmak said, but deferring the vertical problem until AGI can answer it misses the work. Millions of customers bring different accents, contexts, concerns, and re-engagement needs; success requires models, orchestration and harnesses, and purpose-built software working together.

• Sarah observed that founders now avoid verticals they imagine are on a lab roadmap. Tokmak’s sharper diagnosis was short-termism: too much building is oriented around “how can I exit immediately,” whereas founding a consequential company means committing “your whole life” to it for decades.

4. Netic hires for sustained agency and craftsmanship

• Tokmak rejects the “permanent underclass mentality” that says someone must make money within 18 months or learn everything within six months or remain permanently poor. She pairs that with an “AGI-pilled” fear that AI will subsume many skills within 18 months and reduce people’s value. Her counterclaim is experiential: “building really good things takes a very long time,” and the deepest lessons come from staying through successive versions and problems.

• The cultural metaphor she attributes to Martin Luther is the Christian shoemaker who honors God not by adding crosses, but “by building the best shoe.” Applied to Netic, craftsmanship means solving problems for people the team cares about, with focus and a small group capable of “taking down the whole world.”

• Her agency screen follows a person’s whole story: what did they initiate, why did it matter, and did they keep going? A new graduate needs no formal job history, but a weekend project is insufficient; she wants repeated evidence of taking control and changing the future.

• Elad stopped asking “the hardest thing you’ve ever done” after receiving bad answers; Tokmak defended the question’s range. One new hire cited maintaining a demanding health-and-work regimen every day for over 15 years—unflashy, but evidence of commitment “without getting bored.”

5. Revenue proof is rewriting the enterprise and private-equity pitch

• Tokmak calls “old school” essential services a misconception. Large operators are intensely value-focused and can move quickly: one $500,000 contract she described closed end-to-end in 14 days, after the buyer carefully tested whether the value was real.

• Roofing captures the mix of primal and technical operations. Companies still send door knockers, while Netic can combine inbound demand with satellite data on how hurricanes affect roofs, neighborhood context, and roofing materials—equipping agents for better conversations and identifying whom to approach for net-new revenue.

• Private equity’s old playbook of finding an undiscovered gem, changing the team, creating value, and selling it again is changing because “those gems don’t really exist anymore.” Tokmak sees more operating partners and AI-capable engineers, but warns against treating every AI product like software that can be judged after one week: the first week should begin the relationship, and results should persist and improve throughout the year. More deterministic products may still be testable quickly.

• Tokmak estimated Netic has produced “over $600 million” for customers through AI-handled interactions, and she prefers showing live deployments to demos. PE buyers still begin with cost cutting, so Netic must expand the frame: efficiencies may occur, but the central case is measurable new revenue and better service.