Every AI sales tool hallucinates. CortexOS doesn't.
_{A 100% grounded, verifiable AI GTM platform with zero hallucinated facts. Replace your SDR team with an autonomous swarm that actually tells the truth.}

---

⚡ What is CortexOS?

The current generation of AI SDRs and sales tools all suffer from the same fatal flaw: they make things up. They generate company profiles with hallucinated technologies, assign fake titles to real people, and output arbitrary "Confidence: 99%" scores that have no basis in reality. When your outreach is based on a hallucination, you burn your domain reputation and look like a spammer.

CortexOS is built on a fundamentally different premise: If an LLM says it, it must be verified against a source.

It deploys a swarm of specialized AI agents that work in concert to:

1. Discover target accounts from any data source.
2. Research & Extract claims via deep web scraping.
3. Verify every single claim against source HTML using a deterministic Rust pipeline.
4. Score every account against your Ideal Customer Profile (ICP) based only on verified facts.
5. Execute multi-step outreach sequences autonomously.

Zero manual prospecting. Zero copy-paste. Zero hallucinations.

---

🏗️ Architecture & Data Flow

CortexOS operates on a high-performance Rust backend connected to a bleeding-edge React 19 frontend via Tauri's IPC bridge.

---

🛡️ The Verification Engine

Instead of letting an LLM write a summary from its latent knowledge, CortexOS runs a 3-Tier Verification Gauntlet:

• The Gauntlet: Claims extracted by the LLM are passed into a deterministic Rust engine which verifies them against the raw HTML of the source. It assigns a confidence tier based on the matching strictly: Exact Match (100%), Normalized Match (92%), and Fuzzy LCS (>80%).
• Corroboration Engine: O(N²) deduplication merges identical claims found across independent sources (e.g., Tech Stack Agent and Business Model Agent both find the same metric). The UI displays a striking "Confirmed by N sources" pill.
• The Trust Score: Every generated profile is stamped with a deterministic Trust Score—the exact percentage of claims that were mathematically verified.
• The Evidence Tab: In the UI, click on any generated claim to see exactly where it came from, highlighted on the original source URL.

---

🤖 Agent Swarm

CortexOS deploys 8 autonomous agents, each with specialized worker pools coordinated by a central orchestrator.

graph TD
    subgraph Orchestrator["🧠 CORTEX ORCHESTRATOR"]
        direction TB
        O[Central Command]
    end

    subgraph Research["🔍 RESEARCH AGENT"]
        R1[Tech Stack Specialist]
        R2[Business Model Specialist]
        R3[Pain Point Analyst]
        R4[Recent Triggers Spider]
    end

    subgraph Scoring["📊 SCORING AGENT"]
        S1[Requirement Checker]
        S2[Signifier Calculator]
    end

    subgraph Conversation["💬 CONVERSATION AGENT"]
        C1[Topic Generator]
        C2[Personalization Engine]
    end

    subgraph Outreach["📧 OUTREACH AGENT"]
        OA1[SMTP Dispatcher]
        OA2[Reply Classifier]
        OA3[Calendar Coordinator]
    end

    subgraph Signals["📡 INTENT RADAR"]
        SIG1[Job Board Spider]
        SIG2[Bombora Intent API]
        SIG3[LinkedIn Monitor]
    end

    subgraph Persona["🎭 PERSONA MAPPER"]
        P1[Org Chart Builder]
        P2[Role Classifier]
    end

    subgraph ICP["🎯 ICP OPTIMIZER"]
        I1[Neural Feedback Loop]
    end

    O --> Research
    O --> Scoring
    O --> Conversation
    O --> Outreach
    O --> Signals
    O --> Persona
    O --> ICP

    Research -->|Verified Facts| Scoring
    Scoring -->|Qualified Leads| Conversation
    Conversation -->|Drafted Copy| Outreach
    Outreach -->|Deal Outcomes| ICP
    ICP -->|Adjusted Weights| Scoring
    Signals -->|Buying Signals| Scoring
    Persona -->|Committee Data| Conversation

    style Orchestrator fill:#1a1a2e,stroke:#FF5500,color:#FF5500
    style Research fill:#1a1a2e,stroke:#00AEEF,color:#00AEEF
    style Scoring fill:#1a1a2e,stroke:#FFB020,color:#FFB020
    style Conversation fill:#1a1a2e,stroke:#00D084,color:#00D084
    style Outreach fill:#1a1a2e,stroke:#00AEEF,color:#00AEEF
    style Signals fill:#1a1a2e,stroke:#A855F7,color:#A855F7
    style Persona fill:#1a1a2e,stroke:#F43F5E,color:#F43F5E
    style ICP fill:#1a1a2e,stroke:#10B981,color:#10B981

Loading

---

✨ Features

🏢 Verifiable Intelligence Pipeline

• Automated Research — Multi-agent swarm (Tech Stack, Business Model, Pain Points, Triggers) extracts intelligence from raw sources.
• The Evidence Tab — Every company profile surfaces the verbatim quotes, citations, and trust badges proving the facts.
• Multi-Source Corroboration — Visually see when independent sources verify the same intel.

🧠 Autonomous ICP Scoring

• Automated Tiers — Evaluate accounts against strict constraints (must pass) and weighted demand signifiers (Tier 1, 2, 3).
• Self-Learning ICP Optimizer — Neural feedback loop that adjusts scoring weights based on won/lost deals.

👥 Persona Mapping & Messaging

• Buying Committee Visualization — Auto-identifies Champion, Economic Buyer, Blocker, End User.
• Relationship Strength Mapping — Tracks depth of connection via a visual 0-100 heatbar.
• Hyper-Personalization — The Conversation Agent generates personalized outreach based only on verified facts and intent triggers.

🔀 Orchestration Rules Engine

• Visual Flow Builder — Drag-and-drop workflow canvas to compose custom agent pipelines using XYFlow. Fully executable.
• Intent Mesh — Live SVG radar mapping signal density per account (funding, hiring, tech stack changes).

⚡ Tauri Native App

• Rust Performance — Uses Tauri v2 to run completely locally, interfacing with APIs quickly via tokio multi-threading.
• Zero-Latency Database — Embedded SQLite store for instantaneous UI updates without cloud lag.

---

🚀 Web App Views

(Screenshots coming soon - Add your beautiful UI screenshots here to showcase the dashboard, pipeline, intent radar, and evidence tab!)

---

📄 Pages

Route	Page	Description
/dashboard	Command Center	KPI stat cards, sparklines, pipeline funnel, live activity feed
/companies	Companies	Full pipeline table with scoring tiers, Kanban board toggle
/companies/:id	Company Detail	Deep research profile, Evidence tab, Trust Score Ring, people mapped
/contacts	Contacts	List view + Buying Committee visualization with Persona Badges
/signals	Intent Mesh	Radar visualization + live signal feed with trigger actions
/outreach	Outreach Command	Sequence timeline, reply cards, meeting tracking
/agents	Agent Swarm	Deploy agents against targets, stream terminal viewer
/icp	ICP Optimizer	Self-learning feedback loop visualizer, emergent insights
/flow	Flow Builder	Visual drag-and-drop workflow canvas (Fully Executable)
/settings	Settings	LLM Keys, orchestration, email, CRM sync

---

🔄 GTM Execution Workflow

Assemble your pipelines visually using the Cortex Flow Builder, and execute them entirely autonomously.

flowchart LR
    A["🌐 Data Sources"] -->|Scrape & Enrich| B["🔍 Research Pipeline"]
    B -->|Verified Profiles| C["📊 Scoring Engine"]
    C -->|ICP Score + Tier| D{"Score > Threshold?"}
    D -->|❌ Low| E["📋 Nurture Queue"]
    D -->|✅ High| F["🎭 Persona Mapper"]
    F -->|Buying Committee| G["💬 Conversation Agent"]
    G -->|Personalized Copy| H["📧 Outreach Flow"]
    H -->|Sequences| I{"Reply Intent?"}
    I -->|😊 Interested| J["📅 Meeting Booked"]
    I -->|🤔 Objection| K["🔄 Handle & Nurture"]
    I -->|❌ Rejected| L["🎯 ICP Optimizer"]
    J -->|Deal Won/Lost| L
    L -->|Weight Adjustment| C
    
    M["📡 Intent Radar"] -.->|Buying Signals| C

    style A fill:#0a0a1a,stroke:#666,color:#ccc
    style B fill:#001a33,stroke:#00AEEF,color:#00AEEF
    style C fill:#332200,stroke:#FFB020,color:#FFB020
    style D fill:#331100,stroke:#FF5500,color:#FF5500
    style E fill:#1a1a1a,stroke:#666,color:#888
    style F fill:#33001a,stroke:#F43F5E,color:#F43F5E
    style G fill:#00331a,stroke:#00D084,color:#00D084
    style H fill:#1a0033,stroke:#A855F7,color:#A855F7
    style I fill:#331100,stroke:#FF5500,color:#FF5500
    style J fill:#00331a,stroke:#00D084,color:#00D084
    style K fill:#332200,stroke:#FFB020,color:#FFB020
    style L fill:#1a2b22,stroke:#10B981,color:#10B981
    style M fill:#1a0033,stroke:#A855F7,color:#A855F7

Loading

---

💻 Tech Stack

Frontend

• React 19
• Vite
• Tailwind CSS v4
• Zustand (Global State)
• TanStack Query v5 (Data Sync & Cache)
• Motion (Framer) (Animations)
• XYFlow / React (Visual Builder)
• Recharts (Data Visualization)
• D3 / Force-Graph (Memory Graph)

Backend (Tauri Core)

• Rust
• Tauri v2
• Tokio (Async Runtime)
• Rusqlite (Local Database)

Intelligence APIs

• Google Gemini 2.5 Flash (Extractor LLM)
• Tavily (Search & HTML scraping)
• Apollo (People Data)
• Bombora (Intent Signals)

---

🗄️ Database

CortexOS uses a local SQLite database embedded in the Tauri backend for zero-latency operations.

Table	Details
companies	Target accounts, ICP scores, firmographics, and Trust Score metrics
contacts	Buying committee members, personas, and relationship strength
signals	Intent signals (funding, hiring, tech stack changes) tied to companies
outreach	Conversation sequences, generated copy, and reply status
settings	API keys, LLM preferences, and orchestration configurations

---

📂 Project Structure

📂 CortexOS
 ┣ 📂 .github
 ┃ ┗ 📂 assets                  # SVG assets, diagrams, and branding
 ┣ 📂 src-tauri                 # Rust Backend (Tauri)
 ┃ ┣ 📄 Cargo.toml              # Rust dependencies
 ┃ ┣ 📄 build.rs                # Tauri build script
 ┃ ┗ 📂 src
 ┃   ┣ 📄 main.rs               # App entry point
 ┃   ┣ 📄 lib.rs                # Command registry and setup
 ┃   ┣ 📂 agents                # Autonomous agent logic (Researcher, Scorer, etc.)
 ┃   ┣ 📂 commands              # IPC commands exposed to React
 ┃   ┣ 📂 core                  # Core engines (LLM, Tavily, Verification, Flow Executor)
 ┃   ┣ 📂 db                    # SQLite schema, queries, and state management
 ┃   ┣ 📂 hive                  # Swarm orchestration and agent messaging
 ┃   ┗ 📂 prompts               # System prompts and agent instructions
 ┣ 📂 src                       # React Frontend
 ┃ ┣ 📄 App.tsx                 # Routing definition
 ┃ ┣ 📄 globals.css             # Tailwind configuration and design system tokens
 ┃ ┣ 📂 components              # UI Component Library
 ┃ ┃ ┣ 📂 flow                  # XYFlow canvas and nodes
 ┃ ┃ ┣ 📂 ui                    # Reusable primitives (Buttons, Cards, TrustBadges)
 ┃ ┃ ┣ 📂 layout                # Sidebar, MainShell, Command Palette
 ┃ ┃ ┗ 📂 ...                   # Feature-specific components (Outreach, Scoring)
 ┃ ┣ 📂 lib                     # Utilities and State
 ┃ ┃ ┣ 📂 ipc                   # Event bridge and Tauri invoke() wrappers
 ┃ ┃ ┣ 📂 store                 # Zustand state stores
 ┃ ┃ ┗ 📂 sync                  # TanStack query clients and optimisic updates
 ┃ ┗ 📂 pages                   # Application Views (Dashboard, Companies, Signals, etc.)
 ┗ 📄 package.json              # Node dependencies and scripts

---

🗺️ Roadmap

• Phase 1: Core Research Pipeline (Tavily + Extractors)
• Phase 2: React 19 + Tauri 2 Foundation
• Phase 3: The 3-Tier Verification Gauntlet (Rust Engine)
• Phase 4: Intent Signals & Demand Radar
• Phase 5: XYFlow Visual Builder
• Phase 6: Automated Email Dispatch (SMTP/IMAP integrations)
• Phase 7: HubSpot & Salesforce Bi-Directional Sync
• Phase 8: Cloud-Hosted Orchestration (Vercel/AWS)

---