Reference
Most of what is published about FHIR is either specification — precise and enormous — or vendor material. The gap in between is the part practitioners actually need: what the trade-offs are, which decisions are expensive to reverse, and what breaks in production.
These are written from implementation work, not from reading the spec back to you. No signup, no gate.
01 / Agents
What changes when the client reading a record is a language model rather than an application whose behaviour was fixed before you granted it access.
02 / Quality
Measures that actually execute. CQL is the specification language; getting from a published measure to a number you can defend is the part that is rarely written down.
03 / Foundations
The parts of an implementation that are decided early and are expensive to revisit: how the data is shaped, how it gets in, and how the system is arranged.
Advisory
These cover the general case. If you are deciding something specific — an architecture, an agent access model, a measure that has to survive audit — that is what the advisory work is for.