What's ready, what's not, where the boundary lives.
Hands-on technical validation across the 4+1 AI Infrastructure framework. A lab takes a slice of the stack, a single layer, a combination, or the whole model, builds it for real, and renders a verdict on where authority actually sits, scored against the 4+1 model and DAPM.
Labs
Borrow the vendor’s plumbing, not its judgment
I built a retrieval pipeline across a public-cloud data plane and a local box, the compute I keep below the cloud’s managed abstraction, to map where authority actually sits across the 4+1 stack. The economics were the boring part: eighty-four cents, the cloud faster. The finding worth keeping is what the managed path quietly decides for you, and the two questions the lab now knows to ask.
Own the weights, or the platform owns you
Lab one found the Spark loses to the cloud on inference. That verdict held only for commodity base models. The moment you need a custom model you own, the cloud stops selling tokens and starts renting you floors, and the managed path takes something you cannot get back: the weights. Throughout, the box means the compute, serving, and training you keep below the platform’s abstraction instead of ceding them, and the NVIDIA DGX Spark is where this lab draws that line.
The validator determines done, not the loop
Everyone is building agentic loops that let the model decide when it is done and retry until it is satisfied. I built a three-tier escalation on a DGX Spark, gated by a deterministic test harness, and metered every attempt. For the local model, the repair loop added nothing. Escalation gated by the evaluator took the pass rate from 25% to 75%. The authority that determines done is the test, not the model, and not the loop.
You can’t automate a process you haven’t encoded
I handed a frontier model my migration control-plane operating model and let it build against my own production estate. The control plane did not fail where the patterns were owned and encoded. It failed where the model became the author of correctness. The original question was whether a migration could be metered under the model. The better question the run discovered is who may author the patterns, the validators, and the done criteria in an LLM-assisted control plane.