From Systems of Record to Systems of Action

A Fundamental Shift in Software: From Systems of Record to Systems of Action

For the last twenty years, enterprise software has been sold according to the rationale that if you own the system of record, you also own the workflow. LineLeader owned the pipeline and account context, and OrderPort owned the financial and operational ledger. Our clients created durable, recurring revenue, as these systems became deeply embedded in day-to-day execution and were difficult to displace. As a result, the businesses were valued on premium ARR multiples, driven by the stickiness, visibility, and perceived defensibility of that revenue.

The model of core systems of record created extraordinary businesses because enterprise work was organized around a human bottleneck, and data was structured and routed to the decision-maker. Large language models (LLM’s) compress the value chain by combining structured and unstructured inputs, identifying patterns, and increasingly triggering downstream actions without waiting for a person to traverse half a dozen interfaces. The workflow becomes a closed loop, and once insights are instantaneous, the scarce asset is no longer the ability to surface information but the ability to act.

Market leaders are already taking the next step. Salesforce, the archetypal system-of-record company, is now explicitly repositioning around agentic execution, arguing that re-architecting enterprise software is not about layering AI on top of existing workflows but about fundamentally changing the value layer of the application. Agentforce is Salesforce's commercial bet on that thesis, and it's accelerating the move to headless architectures. ServiceNow has been even more direct. Bill McDermott has repeatedly described the platform as a system of action, and at Knowledge 2025 positioned the Now Platform as the orchestration layer to put any AI, agent, or model to work across the enterprise. That language reflects how incumbents are repositioning amid repricing in public markets.

Systems of record, like one of our recent clients, ICANotes, are now positioned at an inflection point. In the case of ICANotes, they became more important as sources of governed, auditable truth. Rather than being replaced by AI, ICANotes and EMRs more broadly are well-positioned to become platforms for the agentic future. As workflows shift from human-driven navigation to model-driven execution, the systems that already contain clinical data, encode the workflow, and enforce auditability are the natural environment for that execution.

The old premium was justified by stickiness: the cost of ripping out a system of record was too high, so the revenue was treated as recurring. The new premium will be justified by a more durable structural position within the closed loop. Systems that already hold governed data, encode the workflow, and enforce auditability are not being bypassed by LLM-native architectures, but are becoming the environment in which they run.

Infrastructure-level trust and the stickiness of systems that end users depend on every day will together command a premium more durable than either ever did alone.

‍