Jesper Lowgren
Designing the Enterprise for Agentic AI
Thought leadership on design, architecture, governance, data backbone, and operating model for intelligent systems that act in the real world.
Jesper Lowgren
I help organisations make sense of agentic AI, and design the structures required to use it responsibly, effectively, and at scale.
My work focuses on five connected domains: design, architecture, governance, data backbone, and operating model. Together they provide the foundations organisations need when intelligent systems move from tools that assist work to systems that actively participate in decisions, processes, and operations.
Five Domains for Governed Agentic Systems
Design
How agency is defined before runtime through intent, boundaries, context, authority, and control.
Architecture
How agentic systems are structured so coordination, assurance, and interoperability can scale.
Governance
How autonomy is bounded, monitored, and made accountable across agents and systems.
Data Backbone
How meaning, policy, provenance, trust, and exchange are carried across the system.
Operating Model
How organisations design the structures, roles, controls, and capabilities needed to run agentic AI at scale.
Signature Concepts
Design Time vs Runtime
Agentic systems are shaped across two domains: design time, where the conditions for agency are defined, and runtime, where behaviour is interpreted, enacted, and controlled.
Procedural Logic vs Declarative Intent
Deterministic software follows prescribed steps, but agentic systems require declared intent, constraints, and outcomes that guide action at runtime.
Governance by Design
In agentic systems, governance is not an overlay — it is a built-in design property that shapes how agents decide, act, and remain accountable.
Boundaries Before Behaviour
Before any agent can behave, its boundaries must be made explicit across intent, authority, policy, scope, meaning, proof, and effects.
Governed Emergence
As agents interact, adapt, and compound each other’s actions, the central enterprise challenge becomes how to govern emergence without suppressing useful autonomy.
Shared Semantics
For agents to coordinate effectively across systems and organisations, they must share a common understanding of meaning, context, and intent, not just data formats.
Latest Thinking
Recent essays, frameworks, diagrams, and videos.
Add your first three posts and assign each one a Pillar plus a Format. This section will populate automatically.
Newsletter
Join the Newsletter.
Get clear, structured thinking on agentic systems delivered directly to your inbox.