This past week has been another whirlwind in AI land. From new Agentic AI platforms like ClawdBot/OpenClaw to Moltbook, a social network for agents, and tech-sector sell-offs amid concerns that vendors are indeed delivering on their AI roadmaps and that others could upend their business models.
Last Friday, I presented at an AI summit for manufacturers in Pennsylvania. Earlier that day, we were divided into groups for discussions and to learn from one another. Next to me sat Tom, a mechanical engineer at a 25-person metal fabrication company who’s been learning about (and embracing!) agentic coding assistants for the past year. He shared how he has automated document-intensive finance processes with Claude Code, which he installed on an isolated system. The handful of back-office staff at the company can email Claude Code to trigger the workflow. Tom is the founder’s hero. He increased productivity and automation, won his finance colleague's support, and didn’t break the bank on an otherwise costly application subscription. Tom’s example, together with the recent industry news, gave me enough to think about for the hour-long drive home from the event.
This time last year, a major claim was circulating in the news: “Software-as-a-Service (SaaS) is dead.” According to Satya Nadella (Microsoft’s CEO), business systems such as CRM, ERP, and HCM would become the systems of record for business transitions, and the actual business logic and execution would occur via AI agents. But the shockwaves, if any, were modest. In fact, the share prices of all major software vendors continued to climb. But when OpenAI rival Anthropic recently released new functionality targeting the financial and legal industries, sentiment shifted back into “SaaS is Dead.” But is that really the case?
A Brief History of Software
Several decades before vibecoding applications with AI coding assistants was a thing, a new type of software emerged: standard software. Until that point, companies that had sufficient budget and experienced programmers developed their own applications to run key business operations (or processes). Projects frequently ran over time, over budget, or fell short on quality, and one company’s systems differed from another's.
But closing the books at the end of the quarter or handling debits and credits follows the same accounting standards and industry best practices. So why develop the software that handles these tasks yourself—from scratch? Why not build it once and make it useful for many customers in many industries? It’s why we have a handful of operating systems (Windows, macOS, Linux, Unix, …), financial systems, productivity applications, and so on, rather than every company developing its own.
Over the years, architectures have evolved from mainframe systems to client-server/on-premises to the Cloud. Along with architectural changes, monetization models have evolved, shifting from licenses to subscriptions, for example. However, the idea of using software to automate business processes and tasks remains the same. Some of the largest businesses on the planet still have requirements to run their business systems in their own data centers (10-15 years since Cloud Computing and SaaS became a software delivery model!). System performance, latency, data privacy, and functional concerns are some of the drivers to keep systems “on-premise.”
SaaS is NOT Suddenly Dead, But Software Keeps Evolving
If we’ve seen anything about AI’s developments in recent weeks, it’s again the pace at which innovation is happening. Moltbook, the social network for AI agents, offered a glimpse of a future in which agents communicate and self-organize. What is a Reddit-style forum for our human amusement today can quickly evolve into B2B scenarios, such as automating reorder processes or negotiating deals between procurement and sales agents of multiple companies.
Naturally, there is hope and promise, as well as deep concern, about what the software and workflows of the future will look like and where that leaves SaaS vendors (and their customers). Let’s face it: “SaaS is Dead” makes for a catchy headline. But the reality is much more nuanced. First, the software industry has a history of “is dead” moments that have typically led to evolution.
How this evolution plays out will largely depend on the following factors:
Enterprise software vendors
Conceive, develop, deliver, and drive adoption of Agentic AI innovation
Create a new business or commercial model supported by the new architecture
Customers
Adapt the organization to adopt new technologies
Connect innovation to historically grown infrastructure and processes (technical debt)
Conduct transformation and change management to bring employees and leaders along on this journey
Technology vendors
Deliver robust, repeatable, and reliable results
Mature technology, governance, and reliability of AI technology/ agents
Let’s go back to Tom’s example for a moment. Yes, it’s absolutely fantastic that he’s automated document processing using AI tools for a fraction of the cost of a full-blown subscription to a finance app. But it’s likely also the beginning of a new kind of technical debt for his company. Who can onboard additional departments, fix issues in the existing workflow, or shut it off if needed? Tom. What happens if Tom is unavailable or leaves the company? You ask someone to build the next tool, or forbid anyone from ever touching Tom’s agentic workflow out of fear it might break. (If you think this only applies to a small business in Pennsylvania, I have worked with Fortune 10 companies that have the exact same problem.)
Despite rapid innovation cycles among AI labs and startups, the evolution of SaaS into widely-adopted agentic architectures will take several years, if not the next decade. The evolution (or death) of one delivery model does not imply the demise of the entire field, and Mark Twain’s famous quote comes to mind: “The reports of my death are greatly exaggerated“ (at least those implying a sudden death).
NEW BOOK — The HUMAN Agentic AI Edge
Organizations are racing to deploy Agentic AI, yet few are ready for the risks that emerge when employees use AI without structure, standards, or oversight.
The HUMAN Agentic AI Edge offers leaders a practical blueprint for building accountable AI-ready teams that consistently produce high-quality results. Drawing on real-world knowledge and insights from interviews with more than 50 AI leaders and experts, Andreas Welsch shows how to combine human judgment with Agentic AI capabilities to achieve the performance many organizations expect but rarely deliver. This book prepares you to shape the next generation of AI-ready teams delivering high-quality results with high accountability.
(Reply to this email to receive the launch day special on February 25!)
When Adoption Meets Business Reality
Business leaders are under pressure to lower operational costs, increase productivity, margins, and revenue, and determine how technologies like AI can accelerate these ambitions. Yet these leaders need to carefully balance innovation and risk. While AI agents are a promising and emerging technology, several areas still require further development to make them enterprise-ready and scalable.
A new workflow, tool, or technology that only works half the time or produces unreliable results creates risks to the business operations that these leaders are ultimately responsible for. The old phrase comes to mind: “No one ever got fired for buying IBM!” That’s why few will adopt novel, unproven tools, or at least pilot them with a limited scope and in a part of the business where negative impacts can be absorbed more easily. After all, enterprise software buying decisions are part innovation and part risk aversion.
Organizational change occurs slowly, especially in large organizations, because innovations must be integrated into existing systems and processes. Think about fully autonomous vehicles. Although they have been on the market for more than 10 years, they promise fewer traffic jams and accidents, and offer more convenience, most of us still drive ourselves. Adoption is happening, but it’s not a big-bang, rip-and-replace change. The same is true for large incumbent organizations that face the daily challenge of technical debt, such as reconciling transactions across several dozen finance systems onboarded during M&A of acquired businesses, rather than consolidating them onto fewer than a handful of systems.
Business operations follow standardized processes and procedures to ensure consistent output and quality. AI agents can handle more complex business tasks that involve greater autonomy and uncertainty. Having a swarm of agents close the books, reconcile transactions, and make payments on your company’s behalf is certainly possible, but it carries risk, incurs additional costs, and the outcome is often uncertain. Taking on this transformation and these risks is what’s slowing many down. In the current phase, AI and agents are often layered on top of existing legacy architectures, rather than rethinking and redesigning processes.
But it is not just customers implementing new software who need to evolve. The maturity of Agentic AI platforms must advance in parallel. Interoperability between isolated, vendor-specific ecosystems through open standards and protocols is an important step. Beyond definitions, the implementation of these standards is critical and will occur over time.
Thanks for reading The AI MEMO! Share this thought-provoking post with your network
AI continues to evolve rapidly, driven by innovations in foundational capabilities, including models and agents, from AI labs and other vendors. The recent increase in scope and capabilities is prompting renewed concerns that SaaS, as a software delivery model, will be superseded by Agentic AI, hence negatively impacting the current business models of incumbent firms. Although the financial markets have been reacting sharply, this change will likely happen over a larger timespan.
Let’s continue the conversation at the Gartner Data & Analytics Summit in Orlando, FL (Mar 09-11, 2026). Register here and use code JOINMEINRLANDO for a $450 discount!
Join my bi-weekly live stream and podcast for leaders and hands-on practitioners. Each episode features a different guest who shares their AI journey and actionable insights. Learn from your peers how you can lead artificial intelligence, generative AI, agentic AI, and automation in business with confidence.
What I think is actually dying is the idea that every business problem needs its own standalone subscription. What's emerging is a shift toward composable, AI-native platforms where the value isn't in the tool itself but in how intelligently it connects to everything else. The companies that'll win will actually make the boundaries between applications irrelevant. That's a much bigger disruption than any single product dying.
What I think is actually dying is the idea that every business problem needs its own standalone subscription. What's emerging is a shift toward composable, AI-native platforms where the value isn't in the tool itself but in how intelligently it connects to everything else. The companies that'll win will actually make the boundaries between applications irrelevant. That's a much bigger disruption than any single product dying.