Technical Documentation

Core Architecture

The Intentio framework is built on a strictly typed, modular pipeline. Below are the formal data models and schemas that enable hyper-personalized generative experiences at scale.

Phase 01

Signals Context Window

The raw behavioral and environmental intake. Captured in real-time from the client session without tracking PII. This serves as the initial trigger for the Persona Agent.

Strict Zod Validation

LLM Contextual Reasoning

signals.schema.ts

{
  "session_id": "string",
  "environment": {
    "weather": "string",
    "local_events": "string[]",
    "device_tier": "string"
  },
  "behavioral": {
    "views": "number",
    "scroll_depth": "number",
    "dwell_time": "number",
    "referral": "string"
  },
  "identity": {
    "loyalty": "string",
    "last_purchase": "string"
  }
}

Live Example Payload

{
  "session_id": "sess_82193",
  "environment": {
    "weather": "rainy",
    "local_events": ["Milan Fashion Week"],
    "device_tier": "high-end-mobile"
  },
  "behavioral": {
    "views": 4,
    "scroll_depth": 0.85,
    "dwell_time": 120,
    "referral": "instagram"
  }
}

Phase 02

Liquid Persona

The psychological twin of the user. Inferred by Gemini 2.0 Flash from the Signals Context Window. This model defines high-level intent and emotional state.

Strict Zod Validation

LLM Contextual Reasoning

persona.schema.ts

interface LiquidPersona {
  identity: { crm_data?; segments?; };
  psychographics: {
    archetype: string;
    emotional_state: string;
    core_values: string[];
    risk_profile: "low" | "medium" | "high";
  };
  behavioral: {
    momentum: "slow" | "steady" | "fast";
    friction_level: number;
    engagement_depth: number;
  };
  intent: {
    primary_goal: string;
    constraints: string[];
    confidence_score: number;
  };
}

Live Example Payload

{
  "archetype": "The Discerning Collector",
  "emotional_state": "exploratory",
  "core_values": ["exclusivity", "authenticity"],
  "primary_goal": "discover_unique_luxury_item",
  "confidence_score": 0.94
}

Phase 03

Harmonized Content Atoms

Granular knowledge blocks extracted from N-sources and prioritized based on the Liquid Persona. Each atom is mapped to a universal schema regardless of source system.

Strict Zod Validation

LLM Contextual Reasoning

content.schema.ts

interface HarmonizedContentAtom {
  id: string; // source:external_id
  source: string;
  type: "product" | "editorial" | "asset" | "social_proof";
  elements: {
    element_type: "text" | "media" | "gallery" | "data_grid";
    role?: string;
    value?: string;
    url?: string;
    data?: Record<string, any>;
  }[];
  relevance: {
    score: number;
    reasoning: string;
  };
}

Live Example Payload

{
  "source": "Aura Blockchain",
  "type": "provenance",
  "elements": [
    { "element_type": "text", "role": "title", "value": "Blockchain Passport" },
    { "element_type": "data_grid", "data": { "origin": "Florence", "year": "2024" } }
  ],
  "relevance": { "score": 0.98, "reasoning": "User values authenticity." }
}

Phase 04

UX Blueprint

The structural orchestration of the experience. The IA Builder Agent maps content atoms to Design System components and sequences them into a narrative flow.

Strict Zod Validation

LLM Contextual Reasoning

blueprint.schema.ts

interface UXBlueprint {
  blueprint_id: string;
  narrative_strategy: {
    theme: string;
    tone: string;
    logic: string;
  };
  sections: {
    component_id: string; // e.g., 'hero-section'
    variant?: string;
    data_mapping: Record<string, any>;
    priority: number;
    reasoning: string;
  }[];
}

Live Example Payload

{
  "narrative_strategy": {
    "theme": "Status & Discovery",
    "tone": "Premium & Exclusive"
  },
  "sections": [
    {
      "component_id": "hero-section",
      "data_mapping": { "title": "Reserved for You" },
      "priority": 100
    }
  ]
}

Ready to see the compiler?

Launch Experience Engine