Probook Raises $40M from a16z and Sequoia to Build an AI Dispatch Layer for Home Services

Every working day, somewhere between a quarter and two-thirds of the phone calls placed to America's plumbers, electricians, and HVAC shops go unanswered, and the revenue attached to them does not wait politely for a callback. It is not lost to bad marketing. It is not the product of lazy technicians. It evaporates in the gap between the moment a homeowner with a flooded basement dials the first number on the search page and the moment a dispatcher decides who is sent, when, and in what order, because no system in the modern home-services stack was ever built to own that decision end-to-end. \ Probook believes that decision is the most valuable unclaimed software category in the trades. The New York company announced this morning a $40 million round , comprising a $34 million Series A led by Andreessen Horowitz and a $6 million Seed led by Sequoia Capital, with Sequoia also participating in the Series A, to deploy what it calls an AI Operating System for home-service businesses, built around dispatch first and extended into intake, data cleaning, customer messaging, and outbound on a single shared context layer. \ The pitch is unusual for a vertical-AI company in 2026, because it does not begin at the front door. Probook does not claim to answer the phone faster, write better follow-up texts, or coach technicians into larger tickets as its primary act. It claims to take over the operational decision that every other vendor in the category treated as somebody else's problem, the one that determines whether the customer experience is made or broken, and to be measured in jobs run and points of EBITDA added rather than minutes saved. \ The Most Expensive Decision in the Trades The numbers behind Probook's thesis are not exotic, and that is precisely why operators have lived with them for so long. The US home-services market runs to roughly $842 billion in 2026 , spread across more than three hundred thousand small businesses that are overwhelmingly founder-led, under-digitised, and squeezed by a labour market that is no longer producing replacements: the trades now graduate something closer to 0.6 new workers for every skilled tradesperson who retires , which means the constraint on growth is not demand but the operator's ability to convert the demand it already has into completed jobs. \ The structural problem is simple to state and brutal to solve: every system the operator already owns is a system of record, and none of them is a system of execution. The field-service management platform records the job. The CRM records the customer. The accounting package records the invoice. Then a human dispatcher, often the owner's spouse or a single overworked employee, sits in the middle of all of it and makes the call that none of those systems will make, which technician goes to which job, in which order, on which route, with which information, and that decision is made on instinct, on a whiteboard, on a phone that rings while it is already ringing. \ The macro condition that makes this addressable right now is the missed call, the cheapest and most studied failure in the entire sector. Independent operators miss as many as 62 percent of inbound calls during business hours , every missed call carries an average lost value north of a thousand dollars , roughly four in five homeowners simply hire the first business that calls them back , and phone leads convert at a multiple of web-form leads that no amount of paid search can close on its own. These are not statistics that get fixed by adding another point solution, because the problem is that the stack already contains too many point solutions and no single layer accountable for the operational outcome. \ \ Founder-Market Fit, Measured in Six Summers and One Shop \ If any founder has earned the right to make that diagnosis, it is, plausibly, this one. George Eliadis grew up pressure-washing in upstate New York across six summers in the truck with his father, spending two to three hours a day driving between jobs and missing calls he could not hear over a pressure washer, and he did the unfashionable thing for an AI founder in 2026, which was to spend a summer not at a model lab but inside TR Miller, a forty-million-dollar HVAC, plumbing, and electrical shop in Illinois that became Probook's first customer, where he watched the same problem he had lived as a one-truck operator repeat itself at the scale of a real business. \ That sequence matters because it inverts the standard order of operations for the category. Most of the AI vendors who arrived in home services, in Eliadis's own framing, came because the sector looked attractive on a spreadsheet : a large, fragmented, under-software-penetrated market with obvious entry points at the top of the funnel. Probook came at it from the inside, started at the hardest part of the workflow rather than the easiest, and accepted the unglamorous deployment posture that the hard part requires, which is to show up in person, configure the platform alongside the operator's front-line team, and stay on the hook for the outcomes it sells rather than shipping a login and a dashboard. \ The traction is the data point that closed the round, and it is unusual because it is operational rather than promotional. Summers Plumbing, Heating & Cooling, running fourteen locations and two hundred and sixty technicians on the platform, booked 2,542 jobs in its first month on Probook with zero human intervention. Peterman Brothers centralised dispatch across eleven markets and two hundred technicians without adding overhead . Del-Air, an eight-location Florida operator, runs Probook across the stack and treats it as part of the front-line customer-service team rather than a tool the team uses. That is not the profile of a company selling a feature, because feature-led adoption in the trades historically stalls at the pilot; that is the profile of a company whose customers turn it on for dispatch and immediately ask it to run the next workflow. \ \ The Product: One Context Layer, Built From Dispatch Outward Probook deploys against the systems an operator already runs, learns the shop's jobs and technicians and routes, and then makes the dispatch decision autonomously, cleaning each booking before it is assigned, answering every inbound lead with full context, keeping every customer on one text thread with one number from the first touch through the front door, and leaving humans to manage the exceptions rather than the routine. No rip-and-replace. No new system of record. That posture is itself a strategic weapon, because it sidesteps the multi-year implementation cycles that have killed most enterprise software momentum in the trades and lets the platform prove its value against live jobs from the first week rather than the thirteenth month. \ The architectural claim sits underneath the product, and almost nobody covering the funding is going to dwell on it. Probook built dispatch first, and intake, data cleaning, messaging, and outbound came after, only possible because every one of those functions reads from and writes to the same context layer: the system already knows which technician is where, which job is running long, which customer has called twice, which booking is dirty, which route is efficient, and that shared state is the difference between five disconnected tools that each own a slice of the customer and a single platform that owns the operational picture. The voice agent that does not know the dispatch board is guessing. The follow-up tool that does not know the job ran late is sending the wrong message. Probook's wager is that the context layer, not the voice model, is the product. \ Why "Dispatch Automation" Is Not the Full Story Here The phrase "dispatch automation" has already been flattened into a feature label, so it is worth showing what the underlying claim actually means, because the incumbent has a product with almost the same name and a radically different ceiling. ServiceTitan, the category's system of record and a roughly nine-billion-dollar public company since its December 2024 IPO , sells a module called Dispatch Pro, and the published case studies are instructive: the best documented result took a shop from a ten-to-one to a twenty-to-one technician-to-dispatcher ratio , a genuine improvement built on top of a human-in-the-loop workflow that still assumes a dispatcher sits at the centre of the board. \ Probook's customers are reporting technician-to-dispatcher ratios as high as one hundred to one , which is five times the best published outcome of the incumbent's own dispatch product, and the gap is not a function of model quality, because the underlying language and voice models available to every vendor have largely converged. It is a function of build order. A dispatch feature bolted onto a system of record optimises the dispatcher. A platform built around dispatch from the first line of code removes the dispatcher from the routine path and leaves them to handle the exceptions, and that is the difference between a productivity gain and a structural change in how many technicians a single back-office seat can carry. The buyer who understands the distinction has stopped buying a faster dispatcher and started buying a different operating model. \ The Two TAMs: How to Size This Correctly Here is where the analysis gets interesting, because Probook's market can be sized two completely different ways, and the gap between them is the entire investment story. \ Sized conventionally, Probook lives inside the field-service-management software market, the category ServiceTitan anchored on its way to a roughly thirteen-billion-dollar TAM framing at IPO and more than seven hundred million dollars of platform revenue growing in the high twenties year on year . Respectable, growing, and already defended by a public incumbent with a decade of distribution and a balance sheet, which is to say a difficult place for a Series A company to win by selling another software line item priced against a seat count. \ But that is not the market Probook is claiming. Its category framing, an execution layer scored on jobs dispatched and points of margin added rather than seats sold, points at a share of the $842 billion of home-services revenue that flows through the dispatch decision, and at the labour and back-office pool that decision governs, a surface more than an order of magnitude larger than the software budget it superficially belongs to. There is precedent for this exact maneuver outside the trades. Stripe did not compete inside the payment-processing software market when it launched; it took a percentage of transaction volume, a market orders of magnitude larger than the budget line it appeared to sell against. The dispatch workflow today looks structurally similar to online payments in 2010, fragmented across a voice vendor, a texting vendor, a coaching vendor, and an army of human dispatchers stitching them together at the boundary, and Probook is the consolidation of that stack with the same pricing geometry quietly available to it. \ \ The SaaS Playbook: How This Business Actually Compounds For readers building or evaluating vertical SaaS, the Probook model rewards a closer look, because the classic mechanics show up in unfamiliar shapes. \ Pricing. Seat-based pricing is incoherent for an autonomous layer, because there are no dispatchers sitting in it. The natural model is a platform fee plus a share of the outcome, which is the structure the broader agentic market is already converging on as services become priced like software . The strategic consequence is non-trivial: revenue scales with jobs run and margin recovered rather than logins sold, which means the sales conversation begins with a free assessment of the operator's missed-call and dispatch leakage and ends with a number the owner has already validated against the shop's own profit and loss. \ Net revenue retention. Land-and-expand is built into the architecture rather than the sales motion. An operator who turns Probook on for dispatch is one configuration cycle from intake, one from cleaning, one from messaging, and one from outbound, and because all of it reads from the same context layer, each new workflow the platform absorbs expands the surface inside the same customer without new integration risk. Expansion revenue arrives without expansion headcount, which is what best-in-class retention actually requires, and which is the metric that decides whether the Series B prices the company as infrastructure or as application software. \ The moat. Every dispatch decision the platform makes, every dirty booking it cleans, every route it optimises, every exception a human overrides feeds a dispatch-pattern library specific to how real shops actually run, and that library is generated by doing the work in the field, against operator behaviours that do not exist on the public internet, which means a competitor cannot shortcut it with a larger foundation model. Probook's own backer called it a years-old structural moat , and the phrase is doing precise work: the moat is the accumulated context, not the model. \ The round itself fits the model. At a $34 million Series A on top of a $6 million Seed, Probook has reached live deployments across hundreds of locations and three private-equity-backed roll-ups on a fraction of the capital its nearest adjacent comparable consumed. Avoca , the AI voice company that owns the inbound call for the same industry, raised more than $125 million to reach a billion-dollar valuation, and ServiceTitan raised on the order of $1.4 billion across its private life before going public. A capital-efficient, outcome-priced model that reached referenceable operator scale before announcing its Series A reads as a milestone round, not a moonshot. \ \ Competitive Landscape: Everyone Crowded the Front Door Probook enters a market with vendors on every side, almost none of whom does what it does, because almost all of them built where the leads come in rather than where the decisions get made. \ The first and most crowded ring is inbound voice. Avoca is the category's standout, an AI voice agent that answers the missed call, books the job, and has booked more than a billion dollars of work across eight hundred customers, and it is genuinely excellent at the thing it does, which is the front door. Its structural position is also the inverse of Probook's: it owns the inbound conversation and hands the operational decision back to a human dispatcher and whichever system of record the shop runs. Sameday, ServiceAgent, and Netic occupy adjacent inbound-answering positions. The notable detail, the one that frames the entire competitive question, is that Avoca and Probook already share customers such as TurnPoint and Sila , which means the two are not competing to replace each other so much as racing to define which layer the operator treats as the centre of the stack. \ The second ring is the single-function tools further down the funnel: Hatch on texting and follow-up, Rilla on sales coaching through call recording, each owning a real slice of the customer and none of them owning the dispatch board. The third ring is the system of record itself, ServiceTitan and the SMB platforms beneath it such as Housecall Pro and Jobber, broad by definition and built around human-in-the-loop workflows, retrofitting agents onto an architecture that was never designed to remove the human from the operational path. Probook sits in the quadrant the rest of the map left open: deep execution depth on the one workflow that touches every other, sold to operators who have already bought three of the other rings and discovered that none of them talk to each other. \ \ The customer mix is the structural advantage that the funding does not spell out. Probook's named operators include TurnPoint Services, Master Trades Group, Del-Air, Peterman Brothers, and Sila Services , and three of those are private-equity consolidation platforms rather than independent shops, which matters more than any single logo, because it changes what the product is actually being bought to do. \ The Insight Nobody Is Pricing: Dispatch as an EBITDA Lever Inside a Roll-Up This is the part of the analysis that the spreadsheet version of the story misses entirely, and it is the reason the customer list is the most important slide in the deck. \ Private equity has spent the last three years rolling up the trades, and the pace has not slowed: financial buyers now account for roughly half of HVAC transactions , and the add-on volume has kept compounding as platforms acquire the independent shops around them. The economics of that strategy run on a multiple spread that every operator in the space understands and almost no software vendor prices against. A small independent shop changes hands at four to six times EBITDA. A larger shop with recurring revenue trades higher. A platform recapitalisation, the moment a roll-up of scale is sold to the next buyer, clears something closer to seventeen to twenty times EBITDA , and the difference between those numbers is the entire game: a dollar of EBITDA assembled at the bolt-on level and sold at the platform level is worth several times what it cost to acquire. \ Now follow the arithmetic into Probook's product. A platform running two hundred technicians with a handful of dispatchers, rather than the ten or fifteen the old ratio required, does not merely save salary. It removes a fixed cost from the operating model at the exact layer that private equity capitalises on exit, which means every dispatcher seat that Probook makes unnecessary converts almost entirely into EBITDA, and every point of that EBITDA is marked, when the platform sells, at the seventeen-to-twenty-times multiple rather than the four-to-six the underlying shops were bought at. Probook is not selling software into the trades. It is selling a margin-expansion lever to the specific buyer who pays the highest possible price for margin, at the precise moment that margin is worth the most, and that is a positioning a generic field-service tool cannot occupy because it was never built to run the board without the human. The Investor Thesis: Why a16z, Why Sequoia, Why Now The AI market just flunked its ROI exam, and that is Probook's opening. MIT's NANDA initiative found that 95 percent of enterprise generative-AI pilots delivered no measurable profit-and-loss impact despite tens of billions in cumulative spend, with budgets misallocated toward sales and marketing pilots while the highest-return opportunities sat in unglamorous back-office automation. Probook is almost line for line the inverse of the failure profile: workflow-native rather than user-facing, back-office rather than top-of-funnel, vertical-tuned rather than horizontal, and priced against jobs run and margin added rather than productivity vibes. \ The interesting detail is not that two top-tier firms wrote checks, but that two firms applying entirely different lens systems arrived at the same company. David Haber , the a16z general partner who led the Series A, is a fintech-infrastructure investor whose thesis is to lead with software and layer in financial products, and whose track record runs through Plaid and a roster of vertical-software companies that grew into the money their workflows moved. \ Read through Haber's lens, Probook is not a dispatch tool: it is the layer that owns the operator's jobs and, eventually, the operator's money, with embedded payments and working capital sitting one product cycle past the dispatch board. \ Konstantine Buhler , the Sequoia partner who led the Seed and doubled down at the Series A, underwrites a different thesis entirely, the one his firm calls the agent economy : a services market many times larger than the cloud, in which autonomous agents do work that was previously priced as human labour, sold on outcomes rather than seats. \ \ Read through Buhler's lens, Probook is a textbook service-as-software company, replacing the dispatcher with an agent and charging for the job rather than the login. Buhler's own framing of the bet was that most founders building for the trades have never worked in them, and George has , which is a polite way of saying the founder-market fit is the underwriting. \ That two firms reached the same conclusion from a fintech-infrastructure thesis and an agent-economy thesis is a stronger validation than either alone, and it has a practical consequence: the cap table is now structured to price the Series B against either a vertical-SaaS comparable set or an agent-economy comparable set, whichever produces the better framework eighteen months from now. \ What Has to Go Right Honest analysis requires naming the hard parts, and Probook has three worth naming. \ The first is the one-hundred-to-one ratio, a number extraordinary on its face and one that deserves to be tested across many more operators before it is treated as a fixture rather than a best case. A handful of multi-location shops running the platform at that ratio is structurally different from a sector-wide proof, and the honest reading of the current figure is as a snapshot of present-tense product differentiation, not a forecast that every operator who deploys will reach it; the right number for the median shop is almost certainly lower, and the company's job over the next four quarters is to show how much lower. \ The second is the partner-as-competitor risk that runs through the shared-customer relationships. Probook and Avoca already sit inside the same operators, ServiceTitan sells a dispatch product of its own, and the strategic question that lives inside every overlapping-stack relationship is whether today's adjacent vendor is tomorrow's direct competitor once each side decides which layer it intends to own. The mitigation is that the context layer is harder to replicate than a voice agent or a dispatch module, but the mitigation is not that the risk disappears. \ The third is the liability surface attached to autonomous dispatch. When a platform decides, without a human in the loop, which technician is sent to which home and when, it is taking an operational action with real-world consequences for safety, for property, and for the operator's licence and insurance, and outcome-based pricing on a workflow that dispatches human beings into strangers' houses carries an exposure that a recommendation engine does not. Any vendor that takes the action rather than suggesting it inherits a share of the responsibility for the action, and that is a cost of the category, not a flaw in the company. \ Final Thoughts: The Category Question The most valuable enterprise software categories have always been built where the money already moves and nobody thought to put a system of execution. ERP claimed the transaction. CRM claimed the relationship. The system of record claimed the log. The dispatch decision, the operational choice that determines whether the job happens, whether the customer is kept, and whether the margin shows up, has never had a dedicated autonomous layer in an industry that runs on it every hour of every day. \ If Probook's thesis holds, the prize is not a software line item carved out of the field-service-management budget. It is a share of the dispatch-governed slice of an $842 billion market, and inside that the points of EBITDA that private equity capitalises at platform multiples, and inside that the network effects that come from owning the context the workflow generates as it runs. That is a structurally larger opportunity than the conventional software comparable suggests, and it is the opportunity that explains why a fintech-infrastructure partner and an agent-economy partner wrote into the same round on the same day. \ Funding announcements are easy to make and hard to interpret, but this one comes with an unusually clean test: either the one-hundred-to-one ratio holds across the next cohort of operators and the EBITDA shows up on customer boards in the next four quarters, or it does not. In an AI market exhausted by narrative, that kind of falsifiability may be the most valuable feature this announcement actually communicates. \ Don't forget to like and share the story! :::tip Vested Interest Disclosure: HackerNoon has reviewed the report for quality, but the claims herein belong to the author. #DYOR. ::: \