Insights & analysis
Practitioner-grade analysis for enterprise technology leaders: agent-based development, enterprise AI economics, cloud architecture, and DevSecOps.
Series: Modes of Agent-Based Development
A four-part PCAS series on the two operating models for AI in the software development lifecycle: Mode 1 augmentation and Mode 2 agentic loops. Short reads backed by public metrics, with full-length technical companions available on request.
-
Part 1 of 4
Two modes of agent-based development: why augmentation and autonomy will coexist
What Mode 1 and Mode 2 are, the public metrics behind each, and the three forces that hold both in place for the next 20 to 26 months.
-
Part 2 of 4
The Mode 1 playbook: augmentation that holds up in production
Harness design, work-type segmentation, and the build-versus-maintain cost inversion.
-
Part 3 of 4
The Mode 2 playbook: bounded autonomy at machine speed
Orchestration patterns, action tiers, and cost governance for agentic loops.
-
Part 4 of 4
Choosing the mode: a five-variable selection framework
Spec precision, verification surface, training corpus, platform foundation, and risk tier, applied across enterprise domains.
More analysis
Standalone PCAS pieces on enterprise AI economics, model selection, and cloud cost structure. Full versions published on LinkedIn Pulse.
-
Agentic FinOps
One Model to Bill Them All
Fugu Ultra collapses model routing into one endpoint and meters a third token bucket the buyer cannot itemize. What agentic FinOps teams should log first.
-
Great Houses of AI
The Great Houses of AI: A Strategic Guide to Model Selection
A Dune-framed framework for enterprise AI model selection: four vendor houses, the vendor-versus-independent benchmark gap, and 75-85% cost reduction via multi-tier routing.
-
Cloud economics
How AWS raised prices without raising prices
A +32% net effective AWS rate increase decomposed into three pricing levers, and the disciplined-buyer response to each before the producer-price wave hits.