Cortex

Main Functions & Workflow: Cortex's core workflow consists of five steps: 1. Retrieval: Sentries monitor account performance signals continuously; 2. Analysis: Cortex analyzes signals referencing a unified corpus of platform best practices; 3. Decision Core: Generates recommendations based on an outcome-oriented decision protocol (Action / Reason / Impact - A.R.I. protocol); 4. Action: Optimizations are executed automatically or after expert approval; 5. Memory: Retains every decision, winning pattern, and learning for institutional memory.

Four Marketing Disciplines:

1. SEO: Earn visibility for keywords on search engines; 2. GEO: Earn visibility for keywords on AI channels; 3. CRO: Increase site conversion volume and rate; 4. PPC: Search and social paid acquisition. It covers 10 platforms including Google Ads, Amazon Ads, Bing Ads, Meta Ads, ChatGPT, Gemini, Perplexity, Claude, Klaviyo, and TikTok.

Core Features & Advantages:

• Always On: Sentries monitor accounts 24/7, surfacing and routing issues the moment they appear.
• One Engine, Every Channel: Collapses paid, SEO, CRO, and AI search management into one engine—smarter, faster, and cheaper than paying multiple agencies.
• Compounding Learnings: Sees patterns across hundreds of accounts, applying proven wins so clients never pay for mistakes others already made.
• AI Driven, Expert Managed: AI handles day-to-day work and ships optimizations at an unmatched pace, while experts manage decision-making and sign off on high-stakes calls.
• Real-Time Decisions: Compresses marketing decisions from months to minutes, evaluating, approving, and shipping in the same minute an insight surfaces.
• Outcomes Measured: Projects expected outcomes for every optimization and measures actual results against projections for continuous improvement.
• Scale Without Cost: Handles work additional humans would do, allowing marketing footprint to grow while retainer stays flat.
• Source-Cited Reasoning: Every recommendation is grounded in a 1.5M-passage corpus of primary-source documentation, ensuring every decision arrives in three parts: the action, the citation justifying it, and the expected impact.

Corpus & Retrieval System: Every Cortex decision starts by retrieving relevant pages from a 1.4M-passage primary-source corpus. It uses Hybrid BM25 + halfvec(3072) cosine, fused with RRF, reranked by Cohere rerank-v3.5. It covers 142,660 docs across 68 platforms with a p50 retrieval latency of < 100ms. The model can only write what the corpus can defend.

Typical Use Cases:

1. Paid/Organic Overlap Audit: Identifying terms a brand is paying for in Google Ads while already ranking top 3 organically.
2. Structured Data Audit: Identifying missing Schema (e.g., AggregateRating) on product pages and dispatching an optimizer to draft fixes.
3. AI Search Citation Audit: Checking a brand's citation presence on ChatGPT, Perplexity, etc., for buyer-intent queries, analyzing competitor citations, and recommending targeted authority building.

Sentry Categories: Includes search engine crawlers (Googlebot, Bingbot, Applebot), AI training/search bots (GPTBot, ClaudeBot, PerplexityBot), Core Web Vitals (LCP, INP, CLS, TTFB), Ad platforms (Google Ads, Meta Ads, Amazon Ads), and various Structured Data schemas (Product, Organization, Article, FAQPage, etc.), totaling 36 Sentries.