315 | Breaking Analysis | How AI Stacks are Rewriting the Rules of Business - co-authored with George Gilbert & David Floyer Most enterprises think they run deterministic systems. In practice, they run an application jungle held together by probabilistic human interpretation. That “human semantic glue” - exceptions, meanings, approvals, reconciliations - is why the enterprise still lives with delayed truth, conflicting semantics, high coordination cost, and manual recovery. In our latest Breaking Analysis, we argue that AI doesn’t just modernize IT - it rewrites the operating model of the business. Frontier models are the catalyst and the migration engine, but the real shift happens when enterprises build the full AI stack that can coordinate work across silos and take action with confidence: *Deterministic apps remain the substrate - but they stop being the control plane; *A System of Intelligence becomes the real-time truth layer - harmonizing state, context, and policy so agents can operate; *A System of Agency provides the control loop - perceive, reason, decide, act, learn - with guardrails; *A System of Engagement closes the loop - humans in the workflow so the system improves over time. The economic impact is the platform shift is expensive because it replaces more than servers, storage and networking. It replaces the human coordination layer around fragmented applications. Capital expands toward AI factories - while coordination labor falls - and that’s where the 10x productivity conversations start to become attainable. With comments on the new AI operating model and tokens on the P&L from Jeff Clarke's Dell Technologies World keynote. An instructive example of AI ROI from Douglas Schmitt. A nuanced discussion with Anand Eswaran of Veeam Software applied to our AI stack framework. Full Breaking Analysis - link in comments. #AI #EnterpriseSoftware #Agents #DataPlatforms #SystemOfIntelligence #DigitalTwin #OperatingModel #TokenEconomics
So this was an interesting piece to co-author. Once you remake the technology model to support agents that need enterprise-wide context to make decisions that can optimize outcomes, you need to change the operating model. Functional or matrix or network organizations support specialization within a hierarchy that provides coordination. But that same technology platform that supports agents provides more and more of the coordination "glue" that human managers today provide. Agents capture more and more routine expertise, making humans the scarce, floating resource for the high-value, non-routine decisions. The enterprise gets reorganized around end-to-end measurable outcomes. That provides the organizational throughput for efficiency and the flexibility for custom offerings. And that same agent-driven technology platform continually learns. That turns it into a compounding asset: a platform.
Great job guys! One minor point. While large companies have historically achieved lower unit costs through volume and shared resources, that advantage does not automatically extend to AI efficiency. A focused AI startup operating in a narrow vertical can match or exceed those economies within that domain, and reach scale far faster than a large multi-sector company navigating competing priorities. Many analysts, ourselves included, view fast-moving vertical AI startups as among the most credible threats to large incumbents. This discussion, however, is tangential to our core thesis and risks distracting from it.
I am still trying to figure out what “Abundant Intelligence” is…..
Looking forward to this 😀
David , George Gilbert & David Floyer The full report is a sharp, insightful take. AI isn’t just modernizing IT like SaaS did; it’s poised to rip out the real siloed decisions and reconciliation, fundamentally reshaping how enterprises scale and operate. From complexity to advanced productivity!
Read the full research report: https://thecuberesearch.com/315-breaking-analysis-how-ai-stacks-are-rewriting-the-rules-of-business/