Todd Saunders: The Vertical SaaS Operator Betting That AI Will End Vertical SaaS
Lessons from building Broadlume into a Leading Vertical SaaS, Leading to its Exit and the Founding of Dalton Mills
Todd Saunders has spent a decade quietly proving that the most defensible SMB software companies are not the ones with the best code — they are the ones whose founders are most willing to pick up the phone.
At Broadlume, he turned a struggling horizontal ad-tech startup into the dominant vertical SaaS platform for the U.S. independent flooring industry before selling to Cyncly (TA Associates / Genstar) in 2025. He did it via an 8-company roll-up, an 8,000-person Facebook group he personally answered in five seconds, and a conference (FloorCon) that never broke even on ticket sales but built a moat almost impossible to replicate.
Now, with Dalton Mills, he’s making a bold bet: that AI doesn’t make vertical SaaS better — it ends it. Out of stealth in June 2026 with an oversubscribed $9.2M round led by Founder Collective and Matchstick Ventures, Dalton Mills is positioning itself as “Replit for the trades” — a no-code, agentic platform where HVAC operators, roofers, plumbers, and electricians describe their ideal features in plain English and the software builds it.
If Broadlume was the textbook example of how to win vertical SaaS, Dalton is Todd’s argument that the textbook is now obsolete. The pattern across both is a methodology grounded in some hard-won truths about building for SMBs and the trades:
The brand is the moat — not the product. “51% of why people signed up was the Broadlume brand. 49% was the product.”
Domain expertise, not code, is the scarce input. The software industry had it exactly backwards.
Distribution beats sophistication. “Product is table stakes. It’s all about distribution.”
Here’s how he’s applying those lessons to shape the next generation of SMB and trades tech.
The Journey: From Google Ad Tech to an Accidental Pivot Into Flooring
Saunders grew up around small business — his dad has run one his whole life, his brother runs one now. After William & Mary (BBA Marketing, D1 Track & Field), he spent two years at Google (2013–2015) on the SMB team and then on the AdWords Accelerated Growth one.
He left in 2015 to launch AdHawk — “TurboTax for digital advertising,” a horizontal ad-tech tool for SMBs. The thesis came directly from his Google experience:
“Google was preaching, you don’t need agencies, we can help you. And it dawned on me: Google IS the largest agency. They just don’t call it that. The difference is Google’s margins are 90%, and an agency’s margins are 10%.”
AdHawk grew to ~$10M ARR by 2019 with ~80 people, near-profitable, after a Techstars Boulder accelerator and a Founder Collective seed. But the business had a structural retention problem.
The “Let the Data Pick Your Vertical” Moment
In 2019, with Founder Collective pushing him to look harder at who was actually retaining, Saunders ran the cohort cuts. About a third of AdHawk’s revenue was coming from flooring retailers — and those customers retained dramatically better than anyone else.
“In an ad-tech business, the TAM is infinite. We could have sold forever. But retention was low — everyone was promising you lower cost per conversion next month. Even if you got it from $100 to $50, they’d ask why it wasn’t $20.”
The flooring traction was downstream of a channel partner — FloorForce, a small website agency for flooring retailers run by a former flooring-manufacturer executive named John Weller. Weller could walk into any room in the industry and open any door. AdHawk could raise venture money and build enterprise value. Saunders acquired FloorForce in June 2019, ran both businesses in parallel until October, and then pivoted fully into flooring.
The pivot was financially brutal. Revenue dropped from double digit to single digit in millions of ARR. He had to do his first-ever RIF in 2019 — and then a second, more painful one in March 2020, the day the world shut down.
“If we hadn’t pivoted, our business would have gone to zero in COVID. Instead, home services exploded. They were essential. We grew back to more than $10M ARR by 2021 — mostly organic, fueled by the home-improvement boom.”
The Roll-Up Years
Between 2019 and 2023, Broadlume rolled up 8 companies including a CRM (Retail Lead Management), an ERP (built on a 30-year-old COBOL stack), a website platform, and more.
The math, as Saunders explains it, was simple:
“You’re not just buying software. You’re buying relationships with customers AND relationships with manufacturers. You’re buying domain expertise. And there’s a multiple arbitrage: I could buy at 3–4× revenue and I was worth 10× revenue. From the seller’s side, all they’re doing is trading 50–80% of their business for four years of cash plus 1% of mine.”
He underwrote every deal the same way: never re-cut a deal, never kick the founders out, never run the PE chop-shop playbook. Every founder stayed. In an industry of only about a dozen flooring-software companies that “all talk to each other,” reputation discipline wasn’t optional — it was the entire strategy.
In 2021, PSG put in a $60M growth round. In December 2024, Cyncly (owned by TA Associates and Genstar) acquired Broadlume.
The Bespoke-Software Bet
Now at Dalton Mills (June 2026), Saunders is co-founding with Sean Bave (and an early Head of Product, Connor Kuehnle, ex-Housecall Pro) and building in New York.
In his June 2026 manifesto “The Trades Are Building”, Todd puts the Dalton thesis this way:
“The software industry spent decades treating your knowledge as the cheap part and the code as the expensive part. They had it exactly backwards… AI does not make that knowledge worthless. It makes it the most valuable thing you own.”
If you zoom out, the through-line is consistent: take a market the Valley doesn’t want to talk to, build a community before you build the product, and let the operators tell you what to ship.
SMB Market: Insights and Trends
Where AI Is Real, and Where It Is “Twitter Hype”
Asked what most SMB founders get wrong about AI, Saunders doesn’t hedge:
“In vertical SaaS, these AI go-to-market tools are Twitter hype. If you’re just a behind-the-screen software company sending a bunch of agents or videos, you’re never going to scale that within an industry.”
When asked how he would have changed at Broadlume if he could have used AI, he feels AI’s real value would have been almost entirely internal:
Onboarding & ticketing: Broadlume had whole teams building thousands of similar flooring websites and answering thousands of similar tickets. “90% of that was the same. It should have been streamlined with AI exponentially.”
PIM / catalog data: “The flooring industry has no single set of data. We took data from every manufacturer and manually moved it in Excel. AI could have done all of that.”
ERP onboarding: ERPs take months to learn because employees memorize which buttons to click. “Why does any employee need to know how to click buttons? You say, ‘I just ordered a box of X for customer Y, I need 1,000 square feet,’ and it should just do it — not click the buttons.”
But for the customer-facing layer, he is famously skeptical as he believes that you need to build real relationships with SMB clients:
“A lot of these big vertical SaaS companies have AI support chat — it’s horrendous. Customers want to hear you. They may ask, ‘How can I sell more flooring?’ They’re not asking about a feature. I don’t want them talking to a chatbot about my help articles.”
The Vertical SaaS Evolution: 1.0 → 1.5 → 3.0
He breaks the market into three eras:
1.0 — Classic Vertical SaaS (Broadlume): “I know something about flooring that you don’t. My workflows are unique and better, and I can speak their language.”
1.5 — “Agentic” Vertical SaaS (today’s hype): “Agentic HVAC, agentic flooring, agentic this, agentic that. Just because you’re agentic in HVAC doesn’t mean anything. Agentic is not a differentiation. It’s like saying ‘digital’ or ‘internet’ at this point — everything is going to be agentic.”
3.0 — Bespoke Software (Dalton’s bet): “AI enables software to be built by the domain expert. If really bespoke software is the future, there’s a chance this is the last software a customer will ever buy.”
The implication is uncomfortable for incumbent vertical SaaS: if the moat was the code, and code is now cheap, the moat is gone. “The vertical-software product is about 99.9% the same across players — website, CRM, ERP, payments, quoting. Differentiation was always GTM and brand, not features.”
The Trades Are Not a Zero-Sum Market
A piece of conventional wisdom Saunders openly rejects: that local trades are cutthroat competitive.
“Your local HVAC guy doesn’t really want more business. He wants more profitability from the same business — because more business means more technicians, more headaches, more quotes. Private equity wants more, more, more. Most local guys are friends with each other. They go to conferences to hang out. They trade ideas. There’s way more demand than labor.”
That insight reframes how he thinks about distribution: in industries like this, referrals are the channel. The right HVAC operator building the right pricing calculator in Dalton will tell their friends — not because of compensation, but because “they want their friends to see it and use it.”
The Bold Prediction: AI Expands the Labor Pool, Not Replaces It
Most AI-in-trades narratives assume labor displacement. Saunders takes the opposite side:
“We are so short of blue-collar labor. I don’t think AI replaces labor — I think it assists labor. With glasses or an AirPod and a model trained on HVAC, a ‘somewhat handy’ person can become certified faster and operate on the job. You’ll see an abundance of labor, not a replacement. And the money’s there — you can be an electrician making $100K in year one.”
Product & GTM Philosophy: Todd’s Core Frameworks
Saunders’ philosophy across Broadlume and Dalton can be distilled into a few sharp frameworks.
Framework #1: Brand > Product (51/49)
Broadlume’s most contrarian principle was Saunders’ rule that “51% of the reason people signed up was the Broadlume brand; 49% was the product.” For a tech audience that is mostly product-first, this is heresy.
His defense: when you sell to small operators with limited time, low feature fluency, and tight communities, the “who” outweighs the “what.”
Almost all of Broadlume’s leads were inbound.
He claims the brand drove an exit premium with Cyncly.
The ROI on community/conference work was, by his own admission, almost impossible to quantify — but it compounded. “The ROI we got, I don’t even think I can quantify.”
📌 Concrete example: Broadlume closed ERP deals with a 2-minute pitch and no demo — “I can’t even demo this well, but if you want a partner for the next 10 years, sign up with us.” The pitch worked because ~4,000–4,500 website customers already trusted them. “If you want to talk features, we’re probably not the right partner.”
Framework #2: Hospitality > Service
The line that defines almost everything Broadlume did externally can be summarized in the philosophy of Danny Meyer: “Service is what you do. Hospitality is how you make people feel.”
If the product isn’t the differentiator, then the focus is on the relationship building and how you can connect with your customers on an emotional level.
Saunders’ personal playbook was a series of things that famously don’t scale:
An 8,000-person flooring Facebook group he personally answered in ~5 seconds.
His cell number given to every customer.
Working in flooring stores to understand the day-to-day.
Listening to ~10,000 retailer phone calls the way other people listen to podcasts.
FloorCon, a warm-location, family-friendly conference (~7,800 attendees) with Tim Tebow and Andrew Yang as speakers — that never broke even on ticket sales.
He’s candid about the failures also: the Facebook group only worked because the retailers were already on Facebook. “We tried LinkedIn. We tried Slack. It only worked where they already were. Lots of small bets compound.”
⚡ Where hospitality shows up (in Ohio): The biggest non-obvious retention play was a twice-a-year bootcamp at a Hyatt in Columbus, OH — teaching flooring retailers how to use ChatGPT and Canva. “Great for retention. Great for NRR. Onboarding is an investment, not overhead.”
Framework #3: Ecosystem vs. All-in-One
When Broadlume started buying companies, Saunders’ instinct was to rebuild everything into one Service Titan-style monolith. He killed that idea fast.
“We’re horrible at that. We don’t have the time, money, energy, or expertise. We kept a 30-year-old COBOL ERP rather than rebuild it. There was essentially no enterprise value we could build in a 3-year window that made rebuilding make sense.”
Instead, he spent the time on connectors and integration points — letting each acquired product run as its own login, but passing data cleanly. “We owned the ecosystem of one brand.”
The same principle drove the eventual rebrand. Broadlume started as FloorForce, then briefly tried a parent-brand/sub-brand architecture (”FloorForce powered by Flooring Stores”), then collapsed everything into a single brand: Broadlume.
“That was a total shitshow. Internally, people didn’t know where they worked. Externally, nobody knew the story. Multiple brands is double the work, or triple, or quadruple. The hardest decision we made — and the right one.”
Retailers didn’t need one product. They needed one relationship — and one brand to attach it to.
Framework #4: Roll-Up Discipline — “Commit or Don’t”
Saunders’ rules for the 8 acquisitions are a clinic in how not to run a PE-style consolidation in an SMB dominated market:
Underwrite for relationships and domain expertise, not just revenue. “You’re buying relationships — with customers, with manufacturers — and you’re buying someone who can sit at the table.”
Never re-cut a deal. The first founder he treated fairly told the next. Trust compounded.
Keep the founder. Every single founder of an acquired company stayed.
Math has to work without heroics. “I could buy at 3–4× and I was worth 10×. The seller trades 50–80% of their business for 4 years of cash plus 1% of mine.”
Integration is the hardest part. “Great deals are won post-integration. Different ICPs, different codebases, no SSO.”
Dalton Mills: The Bet That Vertical SaaS Has One Era Left
If Broadlume was Vertical SaaS 1.0, Dalton Mills is Saunders’ attempt to skip 1.5 entirely and define 3.0.
“Not Another Field Service Product. The Platform the Builders Build On.”
“We’re building this horizontal harness. Every trades business has CRM → lead → quote → invoice → schedule → dispatch → accounting. We build that layer. Then, when an HVAC operator with 3 vans signs up, inference already knows how to adapt the workflow to them. So 98% of the product is built — between us building the harness and inference doing the rest. The last 2% is the domain expertise.”
The architecture is explicitly modeled on Shopify:
The harness (untouchable): CRM, quote, invoice, schedule, dispatch, payments, accounting. Same way Shopify owns the CMS, payments, and shopping cart.
Modular blocks (no-code): ~90% of customer customization happens here. Workflows, fields, quote logic — all editable by talking to an agent.
Custom code (~5–10%): A roofer types “when pitch is steeper than 9/12, add 40% to the labor line” and the agent builds it. No PM, no ticket, no waiting six quarters.
📌 The live-build moment: In June 2026, Saunders went viral on LinkedIn after posting that on a live customer call, Claude transcribed the conversation in the background and built the requested features in real time — handing the customer a working product about 15 minutes after they described it.
“In a 30-minute call, I can ask 15 minutes of questions about their business and their needs, talk 10 minutes about ours, and using that transcript, the last 10 minutes — it’s built. That’s the magic moment on our sales desk.”
It’s also the same psychological play as Broadlume’s no-demo ERP sell — “You listen to me, and you do as I ask” — just compressed from a 10-year partnership pitch into a 30-minute call.
⚡ The other edge — being heard: Large vertical SaaS companies don’t have time to build features that only a small % of customers want. Dalton flips that math. Even if a customer never publishes the feature they just spoke into existence, the act of seeing it built — in the moment, in response to them — creates an emotional relationship the incumbents can’t afford to offer. “They’re not going to build the software for the 20%,” Saunders said of legacy field-service tools. Dalton’s advantage isn’t just a bespoke software. It’s the feeling of being listened to.
The Counterintuitive Tension: Trade Operators, Building Their Own Software
Saunders’ buyers aren’t engineers or product managers. Many still run the business on pen and paper. So how does he expect them to build their own software?
The operator is the domain expert. The roofer knows how a change order moves through her business better than any PM in San Francisco — she’s always been the right person to write the spec. The problem was that the spec had to travel through a product manager, a designer, and an engineer before anything showed up on her screen, and every hop stripped context. Dalton removes the hops.
Code is now conversation. That only works because the interface has changed. “You type one sentence the way you’d explain it to a new hire,” Saunders said. “The roofer doesn’t need to know they’re talking to an agent. The system already knows you’re an HVAC company. It already knows the security, the infrastructure. The customer is just describing what they want — same as they would to a junior employee.”
A few builders per trade is enough. Saunders doesn’t need every operator to build. He needs one HVAC operator who nails scheduling, one roofer who builds a great damage-claim module — and the community carries what they built to the rest. Most operators will never build a thing. They don’t have to.
The Domain-Expertise Marketplace
The most interesting pillar in Dalton’s GTM is the network as a distribution channel. If a customer builds a useful workflow or template, they can publish it to a marketplace:
“If a $10M plumbing operator builds a workflow and publishes it, and someone else uses that template, we take a residual and compensate the person who built it — like Spotify or Twitter.”
But, he’s quick to add, the incentive is almost beside the point:
“From my previous experience, they probably don’t even need the incentive. They want their friends to see it. The power has been in the hands of guys in hoodies building software. Now the power’s in the builders’ hands — and they love to share it.”
It’s the flooring CRM story scaled into a strategy: at Broadlume, one acquired company began as a CRM built from a flooring expert and then sold to hundreds of his industry friends. Dalton is trying to make that pattern the entire go-to-market.
Go-To-Market Playbooks
Broadlume (2015–2025): Community-as-Moat
Broadlume’s playbook for selling into 15,000 U.S. flooring retailers was less about lead-gen than about belonging.
📘 Playbook Highlights:
Acquire your way into the industry. AdHawk had no credibility in flooring. Buying FloorForce — and keeping John Weller as the face — bought instant relationships, manufacturer ties, and the right to be invited into private suites at conferences.
Brand to feel native, not to feel “tech.” Broadlume avoided the “ad company that raised $20M” energy that the industry distrusted. It showed up speaking flooring language, in flooring rooms.
Hospitality > service. FloorCon, the bootcamps, the Facebook group, the cell number — all engineered to make retailers feel like Broadlume cared about their lives, not just their MRR.
Treat onboarding as investment. Twice-a-year retailer bootcamps (ChatGPT, Canva) drove 105% NRR.
Dalton Mills (2026–): Community + Builder-Influencers
The Dalton GTM is the Broadlume playbook with one critical addition: the customers themselves become the distribution.
📘 Playbook Highlights:
Run the community engine as soon as possible. A private “Blue Collar Builders” Facebook group, a podcast/interview series featuring trade operators building their own AI tools — exactly the FloorForce-era playbook, applied to the trades.
Highlight builder-operators publicly. Saunders has personally featured tradespeople who built AI workflows; one electrician profile (around takeoffs) drew VC interest in the operator himself.
Host operators in person. Including at his own home, mirroring the Broadlume principle that “the only un-scalable things worth doing are the ones nobody else will.”
Hire CX before there is a product. Dalton’s head of CX (previously at Wix and Base44) was hired before there were customers. “Our success on onboarding and migration IS the business.”
Build for the 2% who build for the 98%. “It takes two operators in every industry to then build for the rest of the industry and get you distribution.”
Lean into the QA gap with a sandbox + computer-vision agent. Every customer build runs in a sandbox where a Codex/computer-vision agent stress-tests the workflow and surfaces simpler alternatives before it ever goes live.
Channel Strategies That Still Work
✅ Facebook groups & native communities. “It only worked where the retailers already were.” Find where your customers are spending time and engage them there
✅ Referrals over outbound. “Almost all of our leads were inbound.” Your existing customers remain your best acquisition channel because a referral is embedded with the trust between two professionals
✅ Industry conferences — relationship first, transaction second. The FloorCon conference he built (warm location, family-friendly, headline speakers) outperformed any trade-show booth Broadlume ever ran
Practical Advice for SMB Tech Founders
🎯 Lean into inference — or look elsewhere. “Gut-check that you’re leaning into inference and AI, not away from it. If I can replicate your business instantly through inference, I’d look in a different direction.”
📌 Have a right to win that isn’t “I’m smart and the TAM is big.” “If you’re just an MBA student with an idea, a big TAM, and you’re going to build agents — you have no chance. The right to win is brand, relationships, or domain expertise.”
🔍 Distribution over product. “You’re not building robotaxis. You’re not building SpaceX. You’re building a quote-takeoff estimating tool. If you can’t answer simply how you’ll get this into a customer’s hands, I can’t take it seriously.”
🧩 Treat brand as the moat, not the trophy. 51% brand, 49% product. “Almost all our leads were inbound. We got an exit premium for the brand.”
🛠️ Hospitality, not service. “Service is what you do. Hospitality is how you make people feel.” Get your cell number into their phones. Answer the Facebook post in five seconds. Listen to the calls.
📬 Don’t dismiss un-scalable things. “Lots of small bets compound.” FloorCon never broke even — and the acquirer paid a premium for what it represented.
🧠 The founder filter (5 minutes): “Are they sitting behind a computer thinking the most elegant software is their right to win? Or are they working in HVAC companies, listening to phone calls, obsessed with the trade itself? I’d much rather invest in the former technician than the MBA student.”
🚀 The bold prediction: The trades labor pool grows with AI, not shrinks. “AI assists labor — glasses, AirPods, a second brain trained on HVAC. You’ll see an abundance of labor, not replacement. And the money’s there — $100K in year one for an electrician.”
Todd Saunders is Co-Founder & CEO of Dalton Mills, the AI operating system for the trades. He previously founded Broadlume, the leading vertical SaaS platform for the U.S. independent flooring industry, acquired by Cyncly in 2025. He sits on the Westfield, NJ Town Council, was a Forbes 30 Under 30 honoree (2018), and writes about SMBs, the trades, and the future of software on LinkedIn and X.