Sovereignty — Per-Prompt Cartography of Your Maison's Rank
Where Share of AI gives you one number across all prompts, Sovereignty gives you the cartography. Each row of the Battlefield is a prompt monitored weekly: your Maison's rank across the models, the most-cited non-Maison entity on that territory, the worst-case gap to your top tracked competitor, narrative depth, and a Maison-edited strategic status (Sovereign / Conquest / Coverage / Excluded). From any row, four levers — PR Brief, Brand Truth Update, Editorial Outline, Cross-Reference Candidates — generate Comex-grade recommendations targeted at that exact prompt. The drill-down panel shows last-12-runs history, per-model breakdown, action history, and the Brand Truth assertions most relevant to the prompt's category. The territory-by-territory dashboard luxury Maisons have been waiting for.
4 levers
PR Brief · Brand Truth · Editorial · Cross-Reference, per prompt
What you get with Sovereignty
Battlefield — sticky table, sparklines, status dropdown
One row per prompt: average Maison rank, top other citation, top competitor, gap, narrative depth. Sparklines show the last-12-runs trend per row. Status dropdown lets the Maison classify each prompt as Sovereign / Conquest / Coverage / Excluded — persists across runs.
Four levers, fire-and-forget
PR Brief generates 3 angles (Claude Sonnet) tailored to the prompt's gap. Brand Truth Update surfaces the assertions most relevant to the prompt's category. Editorial outlines a 5-section structured brief. Cross-Reference proposes the top Tier 1/2 sources to seed. All fire async; toasts surface completions cross-page.
Drill-down panel with active polling
Click any row, the slide-over opens with 4 tabs: Overview (metrics + assertions), History (last 12 runs sparkline + timeline), Per Model (breakdown across ChatGPT / Claude / Gemini / Perplexity), Actions (lever audit trail). Polls every 10s while a lever is running so the user sees pending → executed → completed without refreshing.
Grammar of Luxury — the editorial vocabulary
Every prompt is mapped to the 6 luxury-native dimensions of AUGMA's editorial doctrine: Heritage, Desirability, Savoir-faire, Exclusivity, Narrative Authority, Consistency. Filter the Battlefield by Grammar dimension to see which prompts test each editorial territory.
How it works
Every Sunday-evening monitor run, the pipeline computes a row per (brand × prompt × model) into prompt_map_entries — Maison rank, top other citation, top competitor, gap, narrative depth, all derived from the same raw responses that feed Share of AI and Hallucination Tracker. Pure-regex entity discovery (no extra Claude calls) plus a curated 95+ luxury-house allowlist + your brand's auto-generated entity_aliases. The four levers run via a worker pattern: lever click queues a row, fire-and-forget POST to /api/lever-worker, worker generates the output (Claude Sonnet for PR Brief, deterministic for the other three), updates the row to completed. The drill-down panel polls the prompt-level state every 10s while anything is in flight.