SaaS Ecosystem Dynamics

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

The "moat" of SaaS is now a death sentence. Data capture has been considered a SaaS moat, but that's changing rapidly. I've consulted with 10+ SMBs this month who are all facing the same issue: "How do we integrate with SaaS tools that don't offer APIs for AI automation?" The answer is increasingly becoming: "Maybe we don't." Here's what I'm seeing in real-time: AI automation is becoming business-critical. Companies aren't just experimenting anymore. They're building operational workflows that depend on AI, and they need their entire stack to play nicely. A new API hierarchy has emerged: • Tier 1: No API = vulnerability • Tier 2: One-way APIs = limited utility • Tier 3: Robust, two-way APIs = competitive advantage SaaS platforms are suddenly vulnerable... Products built in the pre-AI era with limited API access are facing an unexpected threat: Customer Migration. Switching costs matter less than future-proofing... The traditional "it's too hard to switch" defense is weakening as companies prioritize AI-readiness over short-term convenience. The opportunity gap is widening... Airtable is catching many of these customers in the SMB space, while we're seeing entirely new AI-native products emerging to fill the void. The big insight? What was once a protective moat (closed ecosystems) has transformed into a vulnerability (AI-resistance). Smart SaaS companies are racing to retrofit their platforms with AI-friendly APIs, but many won't move fast enough. For buyers, the message is clear: evaluate your SaaS stack not just for what it does today, but for how well it will integrate with the AI-automated workflows of tomorrow. For SaaS founders: your API strategy isn't just a technical decision anymore—it's existential. Is your SaaS stack AI-ready? Or are you already hitting these integration walls?

The “SaaSpocalypse” framing got the direction right and the structure wrong. It’s not that all SaaS dies. But the value is migrating to the edges of the stack and the middle is compressing toward assembly margins. Ryan Waliany at Entrepreneur’s Edge made the original case: software is now following the same arc hardware did in the 1990s. Stan Shih’s SMILE curve described how PC value concentrated at the edges (chip design ~58% margins, brand and distribution ~58% margins) while the middle (Dell, HP, Foxconn) operated at 15-20%. That pattern held for 30 years because the edges owned scarce assets and the middle did necessary but reproducible work. The same curve is now emerging in software: → Far left: chips and compute. NVIDIA, TSMC. Technical scarcity, ecosystem lock-in. → Left-center: frontier models. OpenAI, Anthropic, Google DeepMind. → The trough: horizontal SaaS without proprietary data or workflow embeddedness. CRM, helpdesk, collaboration, e-signature. Anything where the “build it in a week” test is close to passing. → Right-center: distribution and procurement defaults. Microsoft, Oracle, SAP. → Far right: customer relationships and workflow embeddedness. Salesforce, ServiceNow, Adobe, Veeva, Shopify. I plotted ~35 named companies above. A few of the calls will be controversial: — HubSpot in the assembly zone is harsh. SMB-focused horizontal SaaS without a regulated-data moat is exactly what the framework says compresses, but reasonable people would place it one position to the right. — CrowdStrike at the far-right peak treats it as a customer-relationship play. The bull case puts it left-center as a technical/data moat play. I went with right peak because the durable lock-in is platform consolidation, not threat intel. — Atlassian in the “middle with upside” zone, not in the trough, because of workflow embeddedness in software development. But Rovo’s gross margin defense matters — if it breaks, Atlassian slides into the trough. The investment implication isn’t “short SaaS” or “long SaaS.” Both miss. It’s that dispersion within the cohort is going to be enormous, and the market is currently pricing the middle and the edges nearly the same way. Where would you place the names I got wrong? (Personal framework. Not investment advice. Original framing from Ryan Waliany — worth subscribing to his Substack.)

Most product leaders don’t see it yet. The SaaS model is quietly being rewritten. Not by another platform. But by thousands of intelligent, single-purpose AI agents. For decades, software grew by bundling features. All-in-one platforms promised “integration” and “efficiency.” But AI agents don’t need bundles. They connect through reasoning, not APIs. And that changes everything. Here’s how the Great Unbundling is unfolding: 1- Workflows → From central platforms to modular agents. Teams will assemble their own digital stack, task by task. 2- Interfaces → From dashboards to dialogue. Agents will execute through conversation, not clicks. 3- Data models → From ownership to orchestration. Information will move freely across connected agent ecosystems. 4- Pricing → From subscription to performance. Companies will pay for results, not recurring licenses. 5- Integration → From API calls to context sharing. Agents will exchange knowledge, not endpoints. 6- Product design → From control to collaboration. Users will co-create solutions alongside their intelligent agents. 7- Support → From tickets to self-healing systems. Agents will detect and solve issues before escalation. 8- Security → From perimeters to intent verification. Autonomous systems will authenticate purpose, not passwords. 9- SaaS growth → From expansion to fragmentation. Vertical dominance will yield to distributed, specialized intelligence. 10- Strategy → From platforms to ecosystems. Winners will orchestrate agents, not own customers. Because the future isn’t about owning workflows. It’s about enabling intelligence that works across them. The next era of SaaS won’t look like software. It’ll look like collaboration  between humans and agents. ↝ If you want to understand how agent-first design will redefine enterprise software, follow me, Aditya Santhanam, for deeper insights on building intelligent ecosystems. ♻ Share this with a founder still designing platforms when the future is modular.

Google is actively operationalizing agent-based architectures—bringing LLM-powered agents, tool-use orchestration, and multi-step reasoning into mainstream development via platforms like Vertex AI (now called Gemini Enterprise Agent Platform) and emerging agent frameworks. We’re moving from monolithic SaaS applications to composable, agentic systems: Agents as the new abstraction layer: Instead of rigid workflows, agents dynamically plan, call tools (APIs, databases, services), and iterate toward outcomes. RAG + tool use = contextual intelligence: Enterprise data is no longer locked behind dashboards. Agents can retrieve, reason, and act on it in real time. A2A (Agent-to-Agent) protocols: Systems are evolving toward networks of specialized agents collaborating, replacing tightly coupled application stacks. Declarative intent over UI-driven workflows: Users express goals; agents determine execution paths. This fundamentally challenges traditional SaaS economics: High-margin subscription models are vulnerable when domain experts + agent platforms can replicate core functionality at a fraction of the cost. The moat shifts from UI/feature depth to data quality, orchestration layers, and proprietary workflows. “Build vs Buy” decisions tilt toward build (with agents) for many internal enterprise use cases. We’re not at full displacement yet, but the trajectory is clear:SaaS is being unbundled into data + APIs + agents. The winners will be those who: Expose clean, composable interfaces Own high-quality, structured + unstructured data Provide orchestration, governance, and evaluation layers for agents The rest risk becoming thin wrappers around capabilities that agents can reproduce on demand.

Bain & Company just published the clearest framework I've seen for how AI reprices SaaS. Written by Greg Callahan - if you're in enterprise software - operator, investor, board member - this is required reading. The 4-quadrant model (automation potential × penetration potential) gives you a vocabulary to assess where your products sit: stronghold, open door, gold mine, or battleground. The three-layer stack (Systems of Record → Agent OS → Outcome Interfaces) maps where value is migrating. Action Stacks! And the semantic gap they identify - MCP/A2A solve transport but not meaning - is the next strategic frontier nobody's staffed for yet. SaaS down 27% while Nasdaq up 50%. Horizontal SaaS down 49%. This isn't noise. This is the market repricing entire categories in real time. If your leadership team hasn't internalized these dynamics, start here. Block 30 minutes. Read the whole thing. Then ask: which quadrant are we in, and what's our plan? The window to be proactive is closing fast.

SaaS isn’t slowing down. It’s getting eaten alive ... cannibalized. Look at the Aventis Advisors growth chart. The trend isn’t subtle, it’s a cliff. We went from 36 percent growth to 12 percent. Almost a decade of down and to the right. With forecasts pointing to 11 percent … and falling, this feels like a pretty big structural shift. SaaS is starting to look like utilities and pipelines, durable and necessary, but no longer where the real upside lives. And the reason is pretty simple. AI is cannibalizing the very work SaaS used to monetize. Here’s what the chart doesn’t show, but every operator feels. 1) SaaS used to sell “workflows.” AI sells “outcomes” (OaaS). Agents do the work inside the tool, so the tool stops being the product. The labor becomes the product. 2) Budgets (and investors) are leaving SaaS and flowing to digital labor. CFOs aren’t buying more seats. They’re buying fewer humans. AI fits. SaaS doesn’t. 3) Feature parity killed differentiation. Entire categories are indistinguishable. CRM, CX, marketing automation … all the same. AI exposes how thin the moats always were. 4) Enterprises hit peak-SaaS years ago. Now they’re consolidating and cutting 20 to 40 percent of their stack. AI accelerates that purge. 5) AI startups are growing at speeds SaaS can’t touch. When companies hit nine figures in months, not years, investor expectations reset. SaaS looks slow, expensive, and operationally bloated. 6) Value is moving down the stack. The action is in compute, data, agents, and orchestration. SaaS is becoming a UI layer that AI sits on, not the engine driving the work. The growth-rate collapse isn’t a mystery, it’s more of a transfer of value. SaaS is maturing into a stable, cash-flow asset class. AI is becoming the new growth engine of the enterprise. That means founders have a choice, build SaaS and optimize it like infrastructure, or build AI agents that replace the workflows SaaS was built to capture. One path gives you stable multiples, the other gives you growth.

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.

Today, B2B SaaS products perform impressively in isolation, providing functionality, efficiency and productivity gains. But they don’t play well with others. Vendors know they need to offer a wide set of native integrations, but that’s getting harder to achieve. As the B2B tech stack swells (the average business uses 371 SaaS apps), the number of integrations vendors need to build is skyrocketing. In the coming decade, this problem will increase even further as B2B software will operate across thousands of highly specialized applications. These systems won’t just coexist, they’ll need to interoperate in real time, across dynamic, evolving workflows. Current SaaS architectures struggle with integration complexity. Fragmented stacks, ad hoc APIs, and manual workarounds introduce bottlenecks at scale. To fully unlock the value of SaaS, vendors require infrastructure that abstracts the burden of bespoke integration development. Legacy solutions fall short: Embedded iPaaS enables point-to-point connectivity but lacks scalability and maintainability. Unified APIs offer abstraction, but constrain customization and depth of integration due to rigid schemas. What’s needed is a universal, API-agnostic integration layer, one that enables composable, reusable logic across heterogeneous systems at scale with hundreds of apps. At Integration App, we’re building exactly that. Our platform introduces a standardized integration framework that decouples integration logic from underlying APIs. Using AI, we generate adaptive, app- and tenant-specific implementations, allowing developers to build complex, multi-surface integrations with minimal overhead. This architecture dramatically reduces time-to-integration, supports scalable extensibility, and aligns with modern expectations for one-click deployments and dynamic orchestration. SaaS value is shifting from standalone features to ecosystem interoperability. The next generation of platforms will be defined by how well they connect.