Persistent AI Sanctuary

Private & Secure

Home page of You Want It

Step into your phone-first personal intelligence hub. Forge private AI companions backed by Cloud Firestore that remember your habits, store critical memories, and generate bespoke visuals whenever you ask.

Start Free Onboarding Explore Features

Launch User App

Free AccessUnlimited private companions with Firebase Authentication

Quick Navigation

User PortalPersonal workspace & companions

FeaturesMemory bank & visual engine

AboutPlatform vision & architecture

FAQInstant answers & guidelines

ContactFree registration & onboarding

12k+Active Users

500k+AI Memories

85k+Daily Chats

5.0/5

Loved by creators and power users

Companion NodeLuna Personal

Memory Active

Luna

Creative Mentor

MEMORY BANK SYNCEDPRIVATE

Fact Stored

Preference saved to memory

Visual GenNow

Image Generated

Concept art based on your prompt

Infrastructure:Cloud Firestore & Firebase Auth for secure, persistent data.

FIRESTOREFIREBASESECURE AIHTTPS

UWANTIT Community Hub

Live Community Stream

Explore generative notes, private memory snippets, and shared chats created across personal AI companions.

Create & Publish Post

Share updates and memories with your companion

AI Synced

Recent Discoveries

Auto-updating via Cloud Firestore & WordPress

Platform Features & Intelligence

Core Capabilities

Build persistent AI companions that remember your context, craft imagery, and help you achieve your goals.

01Persona Engine

Custom AI Personalities

Design unique AI companions with distinct traits, voices, and behaviors tailored to your specific needs and preferences.

Create companion

02Memory Bank

Persistent Context Recall

A secure memory vault that stores your goals, facts, and preferences, ensuring your AI companion never forgets a detail.

View memory bank

03Visual Synthesis

Instant Image Generation

Bring your ideas to life with integrated image synthesis, allowing your companions to visualize concepts on demand.

Try image tools

04Active Intelligence

Proactive Assistance

Dynamic chat interfaces that evolve with you, providing smart suggestions and task management based on your history.

Start a chat

Secure Data Persistence

All your companions, chats, and memory banks are stored securely in Cloud Firestore and remain private to your account.

Memory Architecture & Context Pipeline

Engineered for Perfect Recall

Discover how UWANTIT transforms conversations into persistent intelligence using ranked memory scoring, a 20-message context buffer, and strict server-side security.

END-TO-END DATA FLOW PIPELINE

30 MEMORIES + 20 MESSAGES

STAGE 01

Memory Ingestion

User prompt and preferences parsed via chat stream.

RAW INPUT

STAGE 02

Importance Tagging

Fact extraction and significance score tagging (0.00–1.00).

SCORING

STAGE 03

Ranked Retrieval

Dual load: top 30 memories (by score) & last 20 messages.

TOP 30 + LAST 20

STAGE 04

Context Window

System payload formed in server memory buffer.

BUFFER PREP

STAGE 05

AI Inference

Isolated server-side execution without client key exposure.

SECURE RESULT

01INGESTION // TAGGING

Extraction & Importance Tagging

Facts are extracted from dialogue and scored for long-term relevance.

When personal facts, goals, or preferences are shared, the system creates an immutable record tagged with an importance score (0.00–1.00) to ensure high-value context is prioritized.

MemorySchema.ts

FIRESTORE

interface MemoryNode {

ownerId: string;

companionId: string;

text: string;

importance: number;

0.00–1.00

createdAt: Timestamp;

}

Automatic WeightingScored On Ingestion

02RETRIEVAL // DUAL-STREAM

Ranked Context Loading

Retrieves top 30 memories by importance and the last 20 recent messages.

Before every completion, the server loads a tailored snapshot: the top 30 memories ranked by importance plus a 20-message conversational buffer to maintain conversational continuity.

context_loader.ts

DUAL QUERY

1. Ranked MemoriesLIMIT 30

where("companionId", "==", id).orderBy("importance", "desc").limit(30)

2. Recent ConversationLIMIT 20

where("companionId", "==", id).orderBy("createdAt", "desc").limit(20)

Index Optimized< 35ms Retrieval

03EXECUTION // ZERO-EXPOSURE

Server-Side AI Inference

Context window assembled server-side with zero exposed API keys.

All retrieval, prompt assembly, and AI model calls happen inside authenticated Cloud Functions. The user client only receives the verified response with zero credential leakage.

inference_gateway.ts

0 KEY LEAKAGE

Frontend Request:auth.token only

Context Assembly:Cloud Function

AI Provider Key:GCP Secret Manager

Zero Client-Side Key Exposure & Owner Isolation

Firestore security rules enforce per-user isolation matching request.auth.uid == resource.data.ownerId. All third-party AI keys reside solely in server secrets.

Firestore Security Architecture

Rigorous Document Ownership & Proxy Protection

Every user action is bounded by Cloud Firestore security rules, server secret API proxies, and strict document-level UID enforcement across all four app collections.

Rules-Level Isolation

Firestore Rules

Document Ownership & 4 Collections

Cloud Firestore partitions data into four distinct collections: users, companions, messages, and memories. Access is gated at the database layer where ownerId must strictly match request.auth.uid.

Target Collectionsusers, companions, messages, memories

Write SecurityStrict UID validation

Read BoundaryZero cross-user exposure

EnforcementownerId == request.auth.uid

Active

Zero PII Exposure

Identity & RBAC

Role-Based Admin Claims

Administrative access is strictly decoupled from end-user records. Custom Firebase Auth claims permit verified owners to inspect registered usernames exclusively, with absolute exclusion of chats, prompts, or emails.

Admin TargetVerified Owner Account

Protected FieldsEmails, prompts, & chats blocked

Claim Validationrequest.auth.token.admin == true

Access ScopeUsernames Only (Custom Claim)

Active

Server Secret Vault

API Protection

Backend Proxy & Zero Frontend Keys

Frontend clients never receive AI provider API keys. All inference and image generation calls flow through an authenticated server-side proxy holding environment secret keys behind HTTPS boundaries.

Key ExposureZero frontend exposure

Auth VerificationFirebase Bearer Token verification

Inference ChannelEncrypted server-to-server TLS 1.3

Client Security0 Client-Side Secrets

Active

Isolated Context Load

Memory Pipeline

Server-Side Context Engine

Prompt assembly happens entirely in memory on the backend. The proxy queries the companion's recent history and semantic bank to load the last 20 messages and top 30 memories per generation cycle.

Recent DialogueLast 20 ordered messages

Memory RecallTop 30 prioritized memories

Payload SafetySanitized before AI dispatch

Context Window20 Messages + 30 Memories

Active

Firestore Rules & Proxy Architecture Matrix

Security rules configuration and server-side context assembly schema

4 Collections Enforced

firestore.rulesSecurity Engine v2

rules_version = '2';
service cloud.firestore {
  match /databases/{database}/documents {
    // 4 core collections: users, companions, messages, memories
    match /{col=users|companions|messages|memories}/{docId} {
      allow read, write: if request.auth != null
        && resource.data.ownerId == request.auth.uid;
    }
    // Admin claim: restricted to username inspection
    match /users/{userId} {
      allow read: if request.auth.token.admin == true;
    }
  }
}

Collections Hierarchy

Separated into /users, /companions, /messages, and /memories with foreign keys keyed to ownerId.

Backend Secret Shielding

AI API tokens remain strictly inside server environment variables. The client communicates solely via authenticated Firebase bearer tokens.

Inference Context Window

Server proxy retrieves exactly the last 20 messages and top 30 memories for the companion, filtering out untrusted input before synthesis.

Enforced Security & Database Integrity

Test live Firebase rule verification and examine end-to-end credential shielding across all database nodes.

Contact Support

4 Collections: users, companions, messages, memories|ownerId == request.auth.uid Enforcement|Server Proxy (20 Msgs + 30 Memories)|Zero Frontend Key Exposure

Join UWANTIT