Systems Thinking

A systems-level view of AI, focusing on architecture, constraints, integration, and operational tradeoffs rather than hype.

Systems Thinking

Developing Agent Personas: when memories should interrupt

Aug 22, 2026

•

10 min read

Developing Agent Personas: when memories should interrupt

Session memory is real value. The harder bet is a personality that grows — and memories that interrupt only when they should.

Stellan Bengtson

Systems Thinking

SpaceXAI + Cursor hints the next moat is ops, not just models

Apr 27, 2026

•

2 min read

SpaceXAI + Cursor hints the next moat is ops, not just models

The real signal in SpaceXAI's relationship with Cursor is not the headline. It is the possibility that the next AI moat is operational: how tightly reasoning, tools, sandboxes, and execution are wired together.What matters here is not just ownership or distribution. It is infrastructure.

Stellan Bengtson

Systems Thinking

The agent market is growing up. Most teams are still building toys.

Mar 31, 2026

•

3 min read

The agent market is growing up. Most teams are still building toys.

What the MCP roadmap says about where agent systems really are in 2026.

Stellan Bengtson

Systems Thinking

Developing Agent Personas: when memories should interrupt

SpaceXAI + Cursor hints the next moat is ops, not just models

The agent market is growing up. Most teams are still building toys.

The Kardashev Space