If you care about Vertical AI, you should probably know Valsoft. Today, Valsoft buys vertical software businesses in the $3M-100M revenue range, runs them in a fully decentralized model, and layers on shared capabilities like global dev centers, payments, and an internal AI lab. The Montreal-based aggregator has quietly grown to 130+ companies, ~3,000 employees across its portfolio, and more than $750M in total revenue by buying, optimizing, compounding mission-critical vertical market software.
Co-founder and CEO Sam Youssef didn’t start out in private markets. Inspired by Warren Buffet’s story after reading The Snowball, he started out trying his hand at value investing… and largely failed. But he then found his way to Constellation Software—founded by longtime VC and fellow Canadian Mark Leonard—which inspired him to build his own vertical compounding machine.
In a moment of growing interest in unique consolidation strategies—from AI roll-ups, to studios, to Constellation-inspired holdcos and micro-PE funds—Valsoft is a fascinating case study… and a window into the future of vertical.
In this episode, we cover: the latest moves in vertical roll-ups and AI agents, how Valsoft actually runs its integration playbooks, and what Sam really thinks about the “death of vertical SaaS” narrative in an AI world.
I) Vertical Market Pulse
The renaissance in roll-up / aggregator models
Beacon raises $250M to buy “Main Street businesses” | Josh Scott @ BetaKit
Beacon Software just raised a $250M Series B at a $1B valuation, led by General Catalyst, Lightspeed, and D1. Co-founders Nilam Ganenthiran (former Instacart president, ex-D1 Capital) and Divya Gupta (former Sequoia partner, ex-Databricks) pitch Beacon as the “anti-private equity firm:” a holding company that acquires profitable, often overlooked, “main street” businesses and equips them with a centralized AI and GTM stack.
Beacon’s model rhymes with Valsoft and Constellation, but with a heavy AI-native emphasis: shared tech, shared AI talent, and a roll-up cadence reportedly as fast as one acquisition every two weeks. We see a few clear signals here:
There is appetite for smaller-scale, profitable vertical businesses
Big platform funds are moving further into historical PE territory
AI is top-of-mind for acquirers, changing how they consider value expansion
Perhaps unsurprisingly given Constellation, Toronto seems to be ground zero for vertical holdco innovation.
AI agents in private wealth & semis
Streetbeat raises $15M on AI for Wealth Managers | Ryan Lawler @ Axios
Streetbeat announced a $15M Series A to scale its agentic AI platform for wealth managers and financial institutions. Its StreetbeatPRO product gives advisors off-the-shelf and custom AI agents that summarize clients, automate workflows, and even construct portfolios, with the company claiming 5x client capacity and up to 15% annual AUM growth for users. Wealth is a classic compelling category in vertical investing given its big, fragmented B2B base and the cash flows it controls. Meanwhile, last-gen innovators like Addepar are coming to maturity.
ChipAgents raises $21M for AI Semi Design | Sally Ward-Foxton @ EE Times
In parallel, ChipAgents closed a $21M Series A to layer agentic AI on top of entrenched EDA incumbents like Cadence, Synopsys, and Siemens, automating RTL verification, debugging, and design optimization for semiconductor teams. There’s been a upswell in the category (e.g. Chipmind, which just raised $2.5M in EU). Often, these require brutally enterprise-y sales motions selling into conservative workflows with stifling long SoR switching cycles—but the prize is huge. Not only because every marginal basis point of chip performance is worth a fortune, but also given obvious recent tailwinds around frontier silicon.
For founders facing entrenched vertical incumbents, there’s a potential learning in both these approaches : find a high-value, deeply technical workflow already sitting on legacy systems of record, and build AI on top of workflows rather than trying to rip and replace. Distribution likely comes via partnerships with incumbents, not by going around them. The question is one we always ask: where do you go from there?
Is loyalty back? B2B2C in restaurants & retail
Magic Raises $10M to Bring AI Into the Real World | Leo Schwartz @ Fortune
Magic announced a $10M seed to bring AI-powered personalization to the “real world,” starting with restaurants via its flagship product Loyalist, an agentic CRM that unifies POS, reservations, private events, and social data into a single guest record. The product sits on top of a restaurant’s existing stack and surfaces “do this now” actions—remembering a guest’s favorite dish, tracking their 10th visit, and triggering the right outreach at the right moment.
Luke and Nic both remember the 2010-era loyalty graveyard of kiosks and punch-card apps that couldn’t get distribution or retention. The difference this time is that tools like Magic plug into existing systems of record instead of trying to replace them, and AI can finally turn messy, multi-system data into something that feels like a real-time memory for staff.
II) Vertical Titan
Sam Youssef — Founder & CEO @ Valsoft
The backstory
In a weird way, Sam’s first business was a roll-up. As a kid, he made cash shoveling snow. It being Canada, after all, he wasn’t the only one. So he saw the chance to “subcontract” out work to friends doing the same. That spark of an idea would resurface in his 20s, when he launched Valsef Capital.
Early on, he tried to copy Buffett’s “value” style by buying cheap public stocks—and got smoked. That pain pushed him away from cigar butts and toward higher-quality, but perhaps still-undervalued companies: sub-scale but durable revenue, mission-critical product focus, proven pricing power, and excellent management teams.
Around that time—about 2012—he discovered Constellation Software, became a big shareholder, and fell in love with the idea of an evergreen vertical market software acquirer. Eventually, he and his partners decided to try their own version—not by raising a giant fund, but by buying a single hotel management software business in South Florida and seeing if they could actually operate it.
That first deal was almost a disaster. The GM quit two weeks after close, leaving Sam—who had never run a software company—to fly down from Montreal, figure out how the business worked, and drag a friend (future COO Michael Assi) into the trenches. Eighteen months later, EBITDA had roughly tripled, and that boutique hotel PMS has since grown into a hospitality portfolio doing $50–60M in revenue and >$10M in profit. A decade later, Valsoft (the flagship, SaaS-focused evolution of Valsef) is nearing $1B in top-line revenue.
The hardest part
Moving from “cheap stocks” to quality
Getting punished buying low-multiple, low-quality names forced Sam to codify what a truly great business looks like: mission critical, recurring, pricing power, and run by high-integrity operators.
Operating the first software roll-up deal
Losing the GM two weeks post-acquisition meant learning everything from product to support to sales the hard way—and doing it while trying not to spook a global base of boutique hotel customers.
Avoiding sub-scale traps
Buying small companies in big markets can be fatal if R&D budgets can’t keep up with better-funded competitors. Valsoft had to learn when “small but mighty” is real (tiny but rational markets) versus when it’s just under-resourced.
Counterparty and customer concentration risk
Some of Valsoft’s worst deals came from low-integrity sellers and businesses with one or two whale customers. When those whales churned, the math broke instantly, regardless of how pretty the deck looked.
Keeping tech debt from killing terminal value
In an AI-accelerated world, legacy stacks that can’t evolve quickly enough will lose customers as those customers themselves fall behind. Valsoft has had to decide when to invest heavily in rewrites—and when to simply pay a lower multiple and accept the risk.
Memorable lines from Sam
“People of high integrity build businesses of high integrity—and vice versa.”
“We run fully decentralized. Our job at HQ is to build an ecosystem where vertical software companies can thrive.”
“I think horizontal solutions are more at risk than vertical systems of record—if those vertical vendors adapt.”
Price matters
Valsoft typically targets businesses between $3M and $100M in revenue to avoid over-concentration in any single asset. From there, pricing is extremely situational: they’ve bought companies for as low as 1x ARR and as high as 10x ARR. The spread reflects product and market reality, not just a negotiation game.
If a product fills a clear gap inside an existing Valsoft vertical—something they can instantly cross-sell into a large installed base—they’re willing to pay up. On the other end of the spectrum, they’ll only touch declining, tech-debt-heavy assets at very low multiples, knowing it may require heavy re-platforming just to get back to neutral. In between are the “good bones” businesses: stable, sticky, but under-optimized, where Valsoft believes its playbooks can unlock faster growth and better profitability.
Small verticals, big advantages
Sam is explicit: the bigger the vertical, the less interesting it usually is for Valsoft. Large markets attract aggressive VC dollars, infrastructure funds, and now AI mega-rounds, which can push competition and pricing into irrational territory.
Smaller verticals—forestry, niche manufacturing, obscure back-office workflow in a number of categories—tend to be calmer. You don’t need a $50M annual R&D budget to maintain a moat and the number of hyper-funded AI upstarts aiming at your customers is much lower. Rational budgets, rational pricing, rational competitors: that is Valsoft’s natural habitat.
Seven integration playbooks
Valsoft doesn’t buy businesses and “let them sit.” Sam talks about seven proprietary integration playbooks they discuss with founders before closing a deal. A few examples:
Embedding a shared payments stack and bundling payments into the core product to expand ARPU.
Injecting 20–25 engineers from global dev centers in Sri Lanka, India, and beyond to burn down tech debt and accelerate feature velocity.
Layering vertical-specific AI features—voice agents, document intelligence, workflow copilots—on top of existing systems of record to capture more of the customer’s workflow and wallet.
The common theme: the more of the workflow you own (without bloating the product), the higher the revenue per customer and the stronger the moat. That’s particularly true now that AI can automate previously manual white-collar work.
The power of decentralization
Every operating business inside Valsoft is run by a GM, often the original founder or a long-tenured vertical expert. HQ doesn’t tell them how to price or which features to ship; instead, it supplies shared services that are too hard or expensive to build locally—dev centers, payments, AI labs, M&A support.
This model also creates an aligned talent market internally: high-performing GMs can graduate to run multiple products or whole operating groups, and best practices diffuse organically as operators compare notes at internal summits and bootcamps. It’s much closer to Berkshire or Constellation than to a top-down-led PE roll-up.
What AI changes for systems of record
Sam takes a clear stance against the “LLMs will kill vertical SaaS” meme. He argues that thin, one-feature products are at risk, but deeply integrated systems of record with strong services, integrations, and embedded vertical expertise are more valuable than ever—if they adapt.
In his view, AI massively increases the amount of technology and automation that can be built on top of a system of record. If your product can’t evolve quickly—because of tech debt or weak teams—your customers will become operationally uncompetitive and will eventually churn. If you can move fast, AI lets you eat more workflow, more revenue, and more margin per customer than the last decade ever allowed.
Hit rate and “learning experiences”
Sam pegs 25–35% of Valsoft’s acquisitions as underperforming relative to expectations. Some are sub-scale in too-competitive markets, some lose key customers, some arrive with more tech debt or cultural debt than diligence surfaced.
But the other side of that barbell is powerful: a meaningful chunk of deals have become 10x+ wins relative to their starting point. That’s the math of a compounding roll-up: you don’t need every deal to work if your upside deals are allowed to keep compounding for decades instead of being sold after five years. A classic VC power-law lesson that applies even for a relatively high-volume aggregator.
When founders should call Valsoft
For founders sitting on profitable, mission-critical vertical software businesses—especially those around $3M+ in revenue with real systems-of-record characteristics—Sam’s advice is simple: build the relationship early. Talk before you’re ready to sell, so both sides can understand the business, the people, and the culture.
When the timing is right, Valsoft can offer founders a way to de-risk personally (via liquidity) while still compounding upside through equity in the broader platform. And because the operating model is decentralized, founders can keep running their baby with more tools, more capital, and a bigger sandbox.
III) Vertical Playbook
Vertical Aggregation
Why it works
The “Valsoft model” combines three compounding engines:
Durable systems of record in rational niches
Mission-critical products with high switching costs and recurring revenue, but in markets too small or boring to attract aggressive, irrational competition.
Buy-and-hold roll-up economics
Acquiring at reasonable multiples, improving growth and margins through repeatable playbooks, and never being forced to sell means value compounds quietly over decades, not through a single exit.
Shared AI and product infrastructure
Global dev centers, a portfolio-wide payments stack, and an internal AI lab drastically reduce the cost and time required to ship new capabilities into dozens of verticals at once.
It works because the incentives line up: founders get liquidity and a long-term home; operators get autonomy and shared tools; and the holdco gets a growing, increasingly diversified stream of cash flows to reinvest into more products and more AI.
How to run it (Valsoft-inspired blueprint)
If you wanted to run a Valsoft-style playbook for M&A or aggregation:
Define your ideal vertical(s)
Fragmented, with many small vendors.
Mission-critical workflows.
Customers with money (or money flows) and willingness to pay for reliability.
Codify acquisition criteria
Revenue range where concentration risk is manageable.
Clear system-of-record characteristics.
Evidence of customer love and low churn.
Technology that can realistically be modernized.
Build integration playbooks before you buy
Know exactly how you’ll add value: payments, new modules, AI agents, cross-sell into a broader portfolio, etc.
Bring those ideas to founders early so they understand what life after close looks like.
Invest in shared infrastructure
Stand up global engineering hubs that portfolio companies can tap into.
Create a shared AI platform for voice, text, and document workflows that each vertical can customize.
Stay decentralized on decisions that touch the customer
Let GMs own product roadmaps, pricing, and customer relationships.
Use HQ to provide capital allocation, M&A support, hiring help, and cross-portfolio knowledge sharing—not to micromanage.
Play infinite games, not finite ones
Avoid leverage structures or fund timelines that force you to sell crown jewels.
Optimize for IRR through compounding, not quick flips.
Founder litmus tests
If you’re a vertical SaaS founder, here are a few questions from this episode’s conversation to sanity-check your own strategy:
Are you the system of record, or just a feature?
If your product disappeared tomorrow, would your customers scramble—or shrug and find a workaround with a spreadsheet and a few AI agents?
Is your R&D set up for the AI decade?
Can you ship AI-powered improvements on top of your current stack quickly, or is tech debt silently capping your terminal value?
Is your strategy aligned with your vertical?
Are you in a weird little niche with limited but rational competition, or in a giant market that attracts mega-round AI entrants who will happily burn against you for years?
What could future M&A look like in your own vertical?
Are there adjacent products you could acquire and embed (like payments, analytics, or adjacent workflows) to expand TAM, or are you boxed into a narrow wedge with adjacent, powerful incumbents?
Would you pass Sam’s integrity test?
If a sophisticated buyer dug under the hood, would they see clean contracts, transparent metrics, and honest disclosure—or a house of cards propped up by one or two big customers?
Whether you’re an investor thinking about the future of Vertical SaaS, a founder considering acquisition / roll-up opportunities, or an entrepreneur envisioning a next-generation aggregator—Sam’s journey at Valsoft offers some excellent lessons.
What we debated on-air
Nic raised the now-classic question: is AI the “death of vertical SaaS,” as some commentators predicted, because end customers will just stitch together agents on top of generic tools? (Admittedly a softball, since we can guess where Nic stands.)
Sam pushed back hard. He sees the risk concentrated in ultra-thin tools, not in deep systems of record that combine software, integrations, and embedded expertise. In his view, AI mostly increases the TAM for incumbents who move quickly, because they can automate more of the customer’s workflow and capture more economics.
Luke pressed on the platform question: should incumbents resist AI upstarts integrating with them, or embrace those integrations as signals of what customers want? Sam’s answer was nuanced—HQ doesn’t mandate anything—but he expects every GM to know who’s being funded in their vertical and to treat integration requests as a strategic data point. Sometimes you should build; sometimes you should partner; sometimes you should acquire.
The trio also debated R&D budgets in an AI era. In theory, AI can let strong teams do more with less. In practice, if you’re starting from a brittle legacy stack, you may need a higher R&D budget in the short term just to reach a place where AI starts compounding for you rather than against you.
Next week on VERTICALS
Tune in next week as we cover Vertical FinTech with a truly formidable operator:
Rahul Hampole at ServiceTitan.
If you’re building (or thinking of building) in Vertical AI—or a seasoned founder wrestling with vertical M&A challenges of your own—we’d love to be helpful. Get in touch with the Euclid team in the comments here or on LinkedIn.
