Vertical Integration in SaaS

Notes from the field ✍ Agents are taking two very different paths into the enterprise: ▪️ Horizontal Agent Platforms: Sell the “What” These companies position as: “Build a fleet of agents” or “Your enterprise AI layer.” It’s a noun-first pitch. You are buying “agents” - powerful, general and, in the abstract, quite compelling. The problem? The buyer now has homework: define the use case, find budget, justify ROI. In other words, the hardest part of the sale gets outsourced to the customer. The usual response: - “Cool… what should we use it for?” - “Who owns this?” - “Which budget does this come from?” - “Can you help us design a use case?” Sales turn into co-creation and roadmaps resemble consulting. Most didn't set out to become systems integrators, but that is where gravity pulls them when the use case must be invented alongside the sale. These platforms are technically powerful but commercially blunt because they lead with capability (agents) instead of pain (a specific broken workflow). ▪️ Vertical Agents: Sell the “Why” They start with: “Reduce support cost per ticket” or “Resolve 60% of IT tickets autonomously.” Now the nouns are irrelevant. Call it an agent, a bot, or magic. What matters is that it attaches to an existing metric and budget. There is an incumbent to displace - no category creation required. Think Decagon in B2C support, Pylon in B2B support, Serval in ITSM. They’re selling outcomes, not AI. The vertical starting point may looks narrower. Increasingly, operators and CTOs are telling a different story: the fastest way to go broad is to start specific and earn your way out. Traditional vertical SaaS gets boxed in by its workflow. AI-native agents don’t, because the core asset is not the workflow but the layer that observes, orchestrates, and accumulates context across systems. Imagine: - A company launches a customer support agent - automating refunds, order changes, subscription issues. Soon they realize most issues are symptoms of pricing and billing friction. Embedded across CRM and billing, it starts triggering fixes, not just answering complaints. Support automation → control layer for customer experience and revenue leakage. - Another launches in IT - password resets, access requests, provisioning. Soon they realize most tickets stem from identity drift. Sitting across HR and IAM, it expands into security (privilege risk, audit) and finance (license optimization). IT automation → control layer for access entropy. Most enterprise workflows are artifacts of how software was purchased, not how work actually happens. You can have different tools across IT, Support, and Security all compensating for the same upstream limitation. Fix the root constraint and you’re not improving a workflow, you’re collapsing artificial boundaries between them. That’s the opportunity. Start vertical to get distribution, trust, and data. Expand horizontally by following the problem, not by declaring a platform.

I've been thinking about vertical SaaS lately. From 2018–21, only 24 % of 80 software IPOs were vertical SaaS. Why? Smaller customer pools and limited value capture kept the upside capped. AI changes the math. Instead of putting clipboards in the cloud, AI does the work itself—and that rewrites three fundamentals: 1. Value | Outputs, not clicks - Pre-AI apps sped up human workflows. - AI-native apps ship the deliverable—draft the brief, reconcile the invoice, triage the patient. When software does the work, it earns a bigger share of the value created. 2. Pricing | Usage, not seats - Seat licenses mapped to headcount. - AI teammates meter documents, calls, or tasks. 3. TAM | Core industry spend, not IT budget - Old ceilings: field-service software ≈ $5.5 B, restaurant POS ≈ $12 B, construction management ≈ $10 B. - New horizon: legal services alone top $1 T. When software augments the lawyer’s, nurse’s, or analyst’s job, it taps the services budget—not just the software line item. Takeaway: Bigger value → usage-based pricing → 100× larger markets. Bonus for founders: Many of these opportunities are untapped.

𝗧𝗵𝗲 𝗹𝗼𝘂𝗱𝗲𝘀𝘁 𝗔𝗜 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻 𝗶𝗻 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝗿𝗶𝗴𝗵𝘁 𝗻𝗼𝘄 𝗶𝘀 𝗻𝗼𝘁 𝘁𝗵𝗲 𝗼𝗻𝗲 𝘆𝗼𝘂 𝗮𝗿𝗲 𝗿𝗲𝗮𝗱𝗶𝗻𝗴 𝗮𝗯𝗼𝘂𝘁. 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 𝗶𝘀 𝗻𝗼 𝗹𝗼𝗻𝗴𝗲𝗿 𝘁𝗵𝗲 𝗺𝗼𝗮𝘁. 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗲𝗱 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗱𝗲𝘀𝗶𝗴𝗻 𝗶𝘀. Across 𝗖𝗲𝗶𝗽𝗮𝗹 𝗖𝗼𝗻𝗻𝗲𝗰𝘁, 𝗣𝗲𝗼𝗽𝗹𝗲 𝗠𝗮𝘁𝘁𝗲𝗿𝘀, panels, one on one conversations, and peer discussions in tech, I have engaged HR and TA leaders from 𝗠𝗮𝗻𝘂𝗳𝗮𝗰𝘁𝘂𝗿𝗶𝗻𝗴, 𝗕𝗮𝗻𝗸𝗶𝗻𝗴, 𝗜𝗻𝘀𝘂𝗿𝗮𝗻𝗰𝗲, 𝗮𝗻𝗱 𝗔𝘂𝘁𝗼𝗺𝗼𝘁𝗶𝘃𝗲. Here is what I keep hearing. ▪ Everyone has AI wins to share. Early use cases, some solid results. ▪ Go deeper into enterprise workflows and the reality shifts. What enterprises need is a foundation of 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗲𝗱 𝗮𝗴𝗲𝗻𝘁𝘀 across the enterprise. Not siloed tools scattered across teams. ▪ Fragmentation is where it breaks. 𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲, 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆, 𝘁𝗿𝘂𝘀𝘁, 𝗮𝗻𝗱 𝗳𝗲𝗮𝗿 𝗼𝗳 𝗴𝗲𝘁𝘁𝗶𝗻𝗴 𝗶𝘁 𝘄𝗿𝗼𝗻𝗴 all surface when there is no unified orchestration. 𝗧𝗵𝗲 𝘁𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 𝗶𝘀 𝗻𝗼𝘁 𝘁𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺. Getting it into real workflows in a connected and governed way is. The ask is consistent. 𝗧𝗵𝗲𝘆 𝗮𝗿𝗲 𝗻𝗼𝘁 𝘀𝗵𝗼𝗽𝗽𝗶𝗻𝗴 𝗳𝗼𝗿 𝗻𝗲𝘄 𝗔𝗜 𝘁𝗼𝗼𝗹𝘀. 𝗧𝗵𝗲𝘆 𝘄𝗮𝗻𝘁 𝘁𝗼 𝗴𝗼 𝗱𝗲𝗲𝗽𝗲𝗿 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀 𝘁𝗵𝗲𝘆 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘁𝗿𝘂𝘀𝘁. Two reports confirm this from the market side. → 𝗥𝗲𝗱𝗽𝗼𝗶𝗻𝘁 𝗩𝗲𝗻𝘁𝘂𝗿𝗲𝘀 (2026 Market Update): 𝗵𝗼𝗿𝗶𝘇𝗼𝗻𝘁𝗮𝗹 𝗦𝗮𝗮𝗦 𝗱𝗼𝘄𝗻 𝟯𝟱%, 𝘃𝗲𝗿𝘁𝗶𝗰𝗮𝗹 𝗦𝗮𝗮𝗦 𝗵𝗼𝗹𝗱𝘀. Vertical platforms own proprietary data, compliance logic, and embedded process history. Switching cost is existential, not cosmetic. → 𝗧𝗮𝗿𝗮𝗻𝗴 𝗦𝗵𝗮𝗵 𝗮𝘁 𝗔𝘁𝗹𝗮𝘀 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 𝗚𝗿𝗼𝘂𝗽 (AI Threat to Software Businesses): systems of record with deep workflow integration are structurally resilient. Prescribed action: 𝗲𝘅𝗽𝗮𝗻𝗱 𝗶𝗻𝘁𝗼 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻. That is exactly what these executives said they are ready to do. Not with new vendors. 𝗪𝗶𝘁𝗵 𝘁𝗵𝗲 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀 𝘁𝗵𝗲𝘆 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘁𝗿𝘂𝘀𝘁. At 𝗖𝗲𝗶𝗽𝗮𝗹, that is the direction we are building in talent acquisition. 𝗗𝗲𝗲𝗽 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝘃𝗲𝗿𝘁𝗶𝗰𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗳𝗶𝗿𝘀𝘁. 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗼𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻 𝗼𝗻 𝘁𝗼𝗽. Trust is earned through depth before intelligence is layered on. The winners in this transformation will be trusted vertical platforms deploying agents across workflows that customers already depend on—𝗡𝗼𝘁 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺𝘀 𝗿𝗮𝗰𝗶𝗻𝗴 𝘁𝗼 𝗮𝗱𝗱 𝗔𝗜 𝗳𝗲𝗮𝘁𝘂𝗿𝗲𝘀. 𝗧𝗿𝘂𝘀𝘁 𝗳𝗶𝗿𝘀𝘁. 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗼𝗻 𝘁𝗼𝗽 𝗼𝗳 𝗶𝘁. Sources: 2026 Market Update by Redpoint Ventures · AI Threat to Software Businesses by Tarang Shah, Atlas Technology Group

The SaaS era was defined by unbundling : find a workflow, optimize it, own it. Salesforce chose sales automation. Slack chose chat. Dropbox chose file sharing. Point solutions won by perfecting single workflows. The playbook : own one pain point, expand from there. AI is moving faster than anyone predicted. When models change every 42 days, buyers can’t assemble a best-of-breed stack. They want a platform they can trust for three to five years. Legal → Professional Services. Harvey now positions itself as AI for legal and professional services, not just law firms. It serves corporate legal departments, court systems, and co-built a Tax AI model with PwC covering 25+ jurisdictions. Enterprise Search → Work AI. Glean started as enterprise search. Now it sells vertical solutions for healthcare, financial services, and government, with dedicated agents for sales, HR, and engineering. Audio Models → Voice Agents. ElevenLabs started with text-to-speech. Now it offers voice agents for customer service, music generation, and AI audiobooks. Foundation model companies are doing the same. OpenAI launched a dedicated Healthcare & Life Sciences vertical, complete with industry-specific sales teams and solutions engineers. Anthropic built an Industries organization with account executives for healthcare, insurance, and federal markets. They’re not selling APIs. They’re becoming platforms. Each of these companies recognized the cognitive burden of unbundling. They’re not selling features. They’re selling trust. There’s a deeper logic at work. Once integrated, AI systems see how teams operate, capture workflows, and build more systems on top of them. As the cost of software development falls, trusted partners with broad adoption can expand faster than anyone else. The SaaS playbook rewarded specialization. The AI playbook rewards breadth.

For decades, vertical SaaS in construction has focused on one thing: Digitizing the Status Quo. We took paper schedules and put them on a screen. We took physical folders and turned them into PDFs. We called it "digital transformation," but really, we were just creating high-tech storage units. The workflow enhancement, the connection of the physical and digital world with complex automation over the last few years that pushed for insights beyond just data, fueled the real digital transformation in the AECO market. AI changes the game because it moves from systems of record to systems of agency. In vertical SaaS—specifically for the AECO industry—AI doesn’t just store your data; it commands it. Here is why AI wins the vertical war: Deep Domain Moats: Horizontal AI (like ChatGPT) is a mile wide and an inch deep. Vertical AI built on proprietary datasets understands the nuance of a "change order" vs. an "RFI" and the legal/financial weight each carries. AI can't replace the why and the certainty that domain brings to the equation. Vertical AI captures institutional knowledge and turns it into a repeatable engine and advanced decision making but in the AECO world. The winners won't just sell software "seats"; they will sell project outcomes. Those outcomes can only be built of the train tracks of deep domain expertise and understanding of complex systems of systems that drive safety and reduce risk and go beyond project management or estimates..... #ConTech #ConstructionAI #VerticalSaaS #DigitalTransformation #AgenticAI

I’ve been watching the SaaS landscape shift and I think we’re about to see the biggest architecture change since the cloud. If we borrow from Sun Tzu, “go where your enemy isn’t”, in SaaS right now means avoiding building yet another monolithic, vertically integrated stack that locks customers in. A traditional vertical SaaS stack owns the workflow end-to-end in a single niche, great for defensibility, but rigid. Legacy SaaS providers are effectively fixed stacks: their integrations, feature set, and data model are tied to a specific platform (such as Shopify), and evolution is slow because every change has to propagate through the entire stack. An agentic SaaS lattice flips that model to something like this: Composable - Instead of one rigid vertical, you have autonomous micro-agents that each handle a specific function (analysis, attribution, segmentation, profitability forecasting, etc.). Evolving - The lattice can reconfigure itself as new data sources, workflows, or priorities emerge... it’s not bound to Shopify, Magento, Salesforce, or any one ecosystem. Feedback Loops Built In - Every agent continuously learns from its own outputs and the downstream results. When the LTV forecasting agent improves, that improvement flows instantly into the budget allocation agent, retention agent, etc. No Platform Lock - Because the lattice connects via APIs and data layers rather than controlling the whole stack, it can be embedded anywhere. The “deep” part is that each agent can reach vertical SaaS depth in its domain, but the lattice as a whole is adaptive and constantly improving. In other words: Depth without lock-in. Adaptability without losing focus. Every agent makes every other agent smarter. The days of the SaaS monolith are over. Generalists will be consumed by specialists. Specialists will link with other specialists (agents) until they match the scale of today’s legacy generalists, but with stronger product-market fit and better unit economics. Monoliths are dinosaurs. AI was the meteor. Agents are what evolved in its wake. It’s already happening… Just last week, Yotpo announced it was exiting its native SMS and Email products, handing them off to Attentive. Instead of stretching to compete in commoditized categories, they’re doubling down on core strengths (Reviews and Loyalty) and partnering with best-in-class specialists for everything else. It’s a shift from trying to be the whole stack to becoming a high-impact node in a smarter, more adaptive lattice. That’s the battleground where the enemy isn’t. In DTC and eCommerce tech, incumbents are stuck in rigid vertical stacks or thin point solutions. Nobody has yet nailed a truly adaptive, feedback-driven agentic lattice for commerce intelligence. Legacy SaaS providers now face a narrowing window to adapt or reinvent. #SaaS #VerticalSaaS #AgenticSaas #SubAgents

One of the biggest software PE firms in the world just called the market wrong. Thoma Bravo held their annual LP meeting last week and shared something tech founders should pay attention to. Quick context: Thoma Bravo is one of the largest software-focused PE firms in the world with $180B+ in AUM. When they talk publicly about market dynamics, it's worth listening. Their core observation: SaaS fundamentals are strong. Public software companies are growing topline at nearly 3x the rate of non-tech S&P 500 companies. Gross margins and revenue durability aren't even close. Still, valuations have compressed sharply – and fundamentals and multiples are moving in opposite directions. Their view on why: the market is overcorrecting. It's repricing software on AI disruption fears that aren't actually showing up in business performance data. This matters for founders for two reasons: One – buyers are active. Strategic and PE buyers see the current dislocation as an opportunity. That's not spin. Thoma Bravo said it explicitly to their LPs. Active buyers with conviction means deal flow and competition for quality assets. Something I can confirm based on my own observations. Two — not all software is in the same boat. Thoma Bravo drew a clear line between businesses at risk and businesses that are insulated. Horizontal tools with simple workflows, commoditized data, and low switching costs → more exposed. Vertical software with deep domain expertise, embedded workflows, and regulatory complexity → more durable. I've been saying something similar about the deals I'm working on. The question buyers are asking isn't "is this a SaaS business?" anymore. It's "how defensible is this specific SaaS business in an AI world?" If your software serves a niche with real complexity and high switching costs, you're probably in better shape than the headlines suggest. If it's more horizontal and commoditized, it might be worth asking whether there's a vertical you could own.

The "SaaSpocalypse" headlines are everywhere. Claude launches workflow plugins. $285 billion in software market cap evaporates in days. CNBC builds a Monday.com clone in an hour for $5. But here's what we may be missing: horizontal AI isn't killing vertical SaaS. It's killing horizontal SaaS, and forcing vertical players to evolve into something far more valuable. Let me explain. Horizontal AI (Claude, ChatGPT, Gemini) is trained on the internet. It knows a little about everything. That's its superpower and its fatal flaw. Ask it to navigate FDA 510(k) submissions while tracking your quality management system, supplier audits, and design history file, and you'll hit its limits fast. The problem isn't intelligence. It's context. Horizontal AI doesn't know a sub-sub-industry's regulatory frameworks, proprietary workflows, the historical data patterns that predict outcomes in a vertical, or the tribal knowledge embedded in how a sector actually operates. This is where vertical SaaS transforms rather than dies. The (surviving + becoming stronger) SaaS companies will use horizontal AI as infrastructure while building defensibility in layers it can't touch. First, they become the action layer, shifting from a system of record to a system of action. They don't just log that a compliance check happened, they perform it. They don't track claims, they adjudicate them. Second, they own the workflow. Data moats are weaker now since LLMs parse unstructured data easily. But workflow ownership is stronger than ever. If you're embedded in the movement of money, atoms, or compliance, you're the execution layer AI needs to function. Third, they build data loops. Static data depreciates. Data loops compound. Every workflow completed improves accuracy, and the product gets smarter in ways specific to your vertical, not general knowledge horizontal AI already has. Fourth, they codify compliance. Regulation is becoming a moat. You can't just feed patient data into generic AI without HIPAA validation. Vertical players who've navigated regulatory complexity for years have infrastructure that's not easily replicated. The revenue model shifts too. Seat-based pricing dies when AI reduces headcount. Outcome-based pricing thrives - charge per claim adjudicated, per report generated, per decision made. Early vertical AI companies are hitting significant percentage of traditional SaaS contract values while growing massively. The "SaaSpocalypse" is real for generic software. For vertical players willing to transform, this is the biggest opportunity in a generation. The question isn't whether to move. It's how fast.

Vertical SaaS is losing the thing that made it vertical. That is hard to say as someone who believes deeply in the category. But it is getting harder to ignore. Much of what is now labeled “vertical software” is built from the same underlying components, with differentiation pushed almost entirely into marketing and onboarding. > A horizontal CRM at the core, relabeled with industry-specific terminology. > An ERP for quoting, accounting, and inventory, differentiated mostly by lingo. > A website platform that integrates with the software but adds no real vertical leverage. > A generic back office for billing, reporting, permissions, and roles. > Off-the-shelf payment rails layered onto software revenue rather than embedded into operating risk. Early on, this worked because owning distribution hid how similar the software really was. If you owned the channel or the community, interchangeability did not matter. That era is ending. The next generation of vertical winners will look less like configurable platforms and more like operating systems for constrained environments. The difference is not UI, terminology, or landing-page jargon. It is responsibility. Old vertical SaaS stopped at abstraction. Next-gen vertical SaaS takes ownership of execution. That means the system does not just store a quote. It decides when a discount is allowed and blocks it when margins break. It does not just track jobs. It schedules work based on real labor availability and reassigns automatically when conditions change. It does not just process payments. It adjusts deposit rules, payout timing, and financing terms as risk increases. This is the fork in the road for vertical software. You can keep shipping tools that describe the business and calling it vertical. Or you can build systems that accept responsibility for outcomes.

The $300B "Correction" was actually a signal. The old SaaS model is expiring. ⏰If you are leading a software company, yesterday was a wake-up call.⏰ The market wiped $300 billion off software and data stocks in a single session. The narrative in the press is fear: Investors worry that new autonomous models will "supplant" traditional software, turning code into a commodity. But if you look past the panic, the market is actually telling us exactly where to go next. It is punishing the Left Side of this forecast and rewarding the Right Side. We are in the middle of a massive transition for ISVs. As I review ISV roadmaps for 2026, here are the three pivots to take action on now: 1️⃣ From Assistive to Agentic (The Product Pivot) Investors are fleeing tools that just "help" users because that value is easily replicated. The future belongs to Agentic AI—systems that don't just chat, but execute autonomous workflows. We need to stop focusing on "productivity" and start driving "work completion." 2️⃣ From Horizontal to Vertical (The Moat Pivot) Broad, "horizontal" SaaS is vulnerable. The new moat is deep, industry-specific data. We must move toward the Vertical SaaS Renaissance—building highly specialized solutions where our proprietary data creates a defense and personalized impact. 3️⃣ From Per-Seat to Outcome-Based (The Revenue Pivot) This is the hardest shift. AI will eventually reduce the human headcount needed to do a job. If your revenue model is tied strictly to "Per-Seat" pricing, your growth is capped. We have to transition to Consumption or Outcome-Based models that capture value based on results, not logins. The sell-off wasn't the end of software; it was the end of the legacy software model. The companies that survive this transition won't just be "systems of record"—they will be systems of action. Which of these three transitions is the biggest risk to your current roadmap? #SaaS #ProductStrategy #AI #Leadership #TechTrends #CloudComputing #GoogleCloud