Layer 1 -- The Panel (six specialists)
HELIOS, NEXUS, ARGOS, VEGA, ATHENA, and PSYCHE each read the week through a single domain lens with boundaries designed to reduce cross-contamination. Each week the scenario is evaluated against the same monitored 40 US-listed ETF universe; the dossier then deep-dives the subset selected for that edition.
Layer 2 -- Five independent SENIOR verdicts
SENIOR instances are architecturally separate pipelines -- same inputs in, five independent conclusions out. The investor-relevant output is not only "what they said" but whether they agree and where they split.
Layer 3 -- PRIME synthesis
PRIME applies a fixed protocol to the five SENIOR outputs: consensus degree, divergence mapping, evidence-quality flags, and systemic bias risk -- producing a week-to-week comparable synthesis rather than a free-form summary.
Why three layers instead of one
Single-layer "one AI answer" products optimize for comfort, not auditability. The three-layer stack trades a fake sense of certainty for structured visibility -- still general information, not advice.
How this connects to your Friday dossier
Every Friday, the dossier you receive contains: the full Panel output (six specialist analyses), all five SENIOR verdicts, the PRIME synthesis, and the mapping of consensus and divergence. You also receive a companion "closure document" -- analysis focused on your portfolio perspective, highlighting relevant entries and exits that week, and long-term reads for positions you plan to hold. It is not portfolio management (that is your decision), but it is portfolio-relevant analysis -- a step beyond generic research. Free subscribers access it 7 days later; Pro subscribers get it immediately plus weekday briefings connecting daily moves to the thesis.
Why the protocol matters over time
Because PRIME applies a fixed protocol every week, you can look back six months and see which lenses flagged what first, which SENIOR verdicts held up, where PRIME's evidence quality warnings proved right. That auditability is how you build trust in a process -- not blind faith in a person or algorithm.