This paper examines where two dominant theoretical frameworks in computational neuroscience — Friston's Free Energy Principle and Tononi's Integrated Information Theory — provide genuine engineering traction and where they stop, drawing on observational evidence from Anima, a running neuroscience-inspired cognitive architecture implemented in Julia. We document specific cases where FEP-derived mechanisms (prediction-error-driven state change, recursive phi feedback, active-inference initiative) produce measurable behavioral effects, and cases where the phi metric as implemented is more decorative than functional. We introduce the concept of endorsement — whether expressed language is consistent with the system's internal state and causal ownership — as a behavioral signal that is more informative than phi alone, and not captured by either framework. No claim is made that either theory "does not work." The claim is that both work differently than their strongest proponents suggest, and that honest engineering requires knowing the difference. Presented as a research artifact with observational evidence from session logs. Not a peer-reviewed experimental result. Project repository: https://github.com/stell2026/Anima