[ The agentic enterprise ]
[ THE MANIFESTO ]
Why enterprise AI in production is a platform problem, and what the stack underneath it actually looks like.
––––––––––––––––––––––––––––––––––––––
01
The Deployment Gap
Every enterprise has agents in pilot. Almost none run them at scale. The gap is not the models, it is everything underneath them.
02
The stack
Four layers: tooling, path specifications, the platform harness, model infrastructure. The reference architecture, embedded live.
03
From IDP to ADP
Platform teams solved this problem once before, for human developers. The discipline transfers; the user changes.
04
The Harness
What changes when the platform's user is an agent: execution, context, capability, evals, and governance as first-class concerns.
05
One substrate, every workflow
Development first, then sales, finance, legal. Why the platform team becomes the enterprise's agentic center of gravity.
06
Measuring what matters
Chips are benchmarked. Models are benchmarked. Workflows are not. The case for measuring ROI where the enterprise actually feels it.
