An early architecture heuristic felt coherent: assign different forms of intelligence to the work they fit best.
A useful synthesis can sound novel when its intellectual ancestry has not yet been excavated.
The framing might carry more originality than the literature supported.
The thesis was subjected to prior-art research across distributed cognition, sociotechnical systems, hybrid intelligence and human oversight.
Important foundations already existed. The stronger contribution was narrower: synthesis, application and open architectural questions for AI-native organizations.
Unexamined novelty weakens credibility and prevents an idea from learning from the work that preceded it.
Giving up an inflated novelty claim made the actual research question more precise.
Prior art is not a threat to the work. It is part of the apparatus for changing the work.
The Architecture of Intelligence v1.1 now names established foundations, EVO observations, synthesis claims, counterarguments and falsification conditions separately.
Ask what belongs to the literature, what is situated observation, and what remains a testable synthesis.
Whether the proposed architecture is useful enough in practice to deserve its own method.