Traditional frameworks (TOGAF, Zachman) were built for predictable roads and rigid zoning. EA 4.0, as Lowgren frames it, is more like SimCity’s sandbox: dynamic, simulated, continuously governed. His nine intelligent systems supply memory, trust, simulation, and execution.
Neuroscience offers a striking parallel: our brains rely on specialised regions as the motor cortex for action, the hippocampus for memory, prefrontal cortex for planning. Yet it’s their synchrony that produces intelligence. Lowgren’s framework recognises that enterprise intelligence needs similar specialisation and integration (akin to Howard Gardner’s multiple intelligences in human cognition).
EA 4.0 in the wild
Leading companies are already treating their architecture as a living system.
- Walmart: An approach that resembles a SimCity-like sandbox: the AI-driven trend-to-product platform synthesises social media trends, customer behaviour, and regional patterns to auto-orchestrate new product development, shrinking the cycle from months to weeks.
- Red Interclínica (Chile): A cloud-based AI analytics hub using BigQuery, Looker, Vertex AI, turning static health records into real-time insights to drive smarter, faster care decisions.
These are EA 4.0 in action: continuously sensing and adapting, rather than following a static blueprint.
Enterprise “brain” architecture
Lowgren suggests nine core systems that appear in every adaptive, agentic enterprise (a sort of enterprise “brain” architecture):
- System of record
- System of intelligence
- System of trust
- System of engagement
- System of collaboration
- System of control
- System of simulation
- System of autonomy
- System of execution
From strategy to execution
In addition, he introduces five agentic forces that reveal how things actually happen. Again, comparing with biology, DNA doesn’t directly build organisms; it flows through RNA to proteins, with epigenetics controlling expression. Lowgren’s five forces mirror this:
- Strategy (DNA) flows through
- Architecture (RNA) and
- Design (tRNA) to
- Operating Model (proteins), all regulated by
- Governance (epigenetics).
Both systems transform information into action through specialised layers; each essential, nonsufficient alone.
From intent to impact: simulate, link, govern
An agentic ontology and capability maps link strategy to code. A system of simulation lets architects stress-test multi-agent behaviour before releasing it to customers. Think of it as a sandbox rehearsal for how a swarm of AI agents might handle a scenario.
As Gurpreet Kaur notes, this is human-in-the-loop by design: humans frame the problem, guide the process, and validate the outcome. The loop isn’t a bottleneck. It’s a learning amplifier.
Does your current enterprise architecture feel more like a rigid blueprint or a dynamic sandbox?
In the next blog post, I’ll share why culture eats AI strategy for breakfast!