Five layers. One throughline. Every signal in real time.

Event Brain · North-Star

Define the event once. Measure everything against it.

Every SYSOI event starts with a North-Star — goals, KPIs, target audience, brand voice, key themes. Draft it from a brief with AI, rebuild it from your website, or paste in your existing strategy doc. Once it's set, every downstream layer is measured against it.

How it works. Opus drafts the North-Star from a brief or scrape, then writes it to a versioned profile. Every signal the Consistency Engine emits, every CFP score Marquee returns, and every readiness model the scoring layer runs reads from this profile. When the North-Star changes — a pivot to a different KPI, an audience refocus — every downstream signal re-scores against the new truth.

Connectors: Web scrape via ForgeScrape; document upload (PDF / Markdown / text); manual brief.

Consistency Engine · Forensic drift signals

A forensic analyst watching your event. Not a chatbot.

SYSOI's Consistency Engine reads the live unified record against your North-Star and emits drift signals across five dimensions — KPI, audience, voice, content, context — with severity ratings, evidence cited inline, and a recommended fix. Every signal is human-reviewed. Every dismissal teaches the system what to ignore.

How it works. Opus runs a forensic scan over a structured snapshot of the event (goals, KPIs, audience composition, session content, brand voice). Outputs are typed signals with severity (info / warn / critical), evidence (the data point that triggered it), and a recommendation. The reviewer queue is human-in-the-loop: acknowledged signals stay open; dismissed signals write to the per-tenant playbook so the same false positive doesn't recur. The Brain-First Protocol weighs prior patterns and prior mistakes before reasoning.

Marquee · CFP fit

Score every session against the event's strategy. With the deck.

Marquee scores every CFP submission, every track topic, and every speaker against the event's North-Star themes, audience ICP, and brand voice — additive math across pillar alignment, audience fit, strategic fit, technical depth, voice fit, and delivery readiness. Upload the actual presentation deck and SYSOI re-scores against what's built, not what's promised.

How it works. Sonnet runs a fixed-additive rubric (six capped factors → 0–100) over each submission. Rationale is the recommendation — accept, accept with revisions, decline. Deck-grounded re-score attaches the PDF as a Claude document and re-evaluates technical depth, voice fit, and delivery readiness against what's actually built. Promote-to-sessions one-clicks accepted submissions into the live agenda the Consistency Engine reads.