Mimic

MIMIC — Privacy-First AI Avatars

Build AI avatars that genuinely reflect your personality, not generic chatbots.

Live Demo: https://mimic-eta.vercel.app/

Render Deployment: https://mimic-zyb3.onrender.com

Video Demo: https://youtube.com/shorts/xD5wGwz724k?si=mCXV4I_gls4C9QjV

WHAT IS MIMIC?

Mimic is a privacy-first AI avatar platform where users create personal AI identities that evolve only through explicit, trusted inputs. Instead of passively learning from every interaction, Mimic uses user-curated memories stored as embeddings and recalled using vector similarity. This ensures accuracy, personality consistency, and minimal hallucination.

THE PROBLEM

Traditional AI chatbots:

Learn passively from every interaction

Cannot be trusted to remember information correctly

Produce generic responses without personality

Give users little control over their data

THE SOLUTION

Mimic provides:

Explicit control: only what you choose gets remembered

Vector-based recall using semantic similarity

Trust weighting to control how strongly memories influence responses

Full transparency showing which memories shaped each reply

Privacy-first architecture with ephemeral conversations and deliberate memory

KEY FEATURES

Smart Memory Add discrete memories such as opinions, preferences, or facts that define your avatar.

Vector Retrieval Memories are stored as embeddings and recalled semantically for accurate context.

Trust Weighting Each memory has a confidence level, allowing high-trust inputs to influence responses more.

Explainability Every response shows exactly which memories were used.

Separate Streams Conversations are temporary. Long-term memory only grows intentionally.

Reduced Hallucination Responses are grounded in trusted, relevant memories.

Scalable Architecture Pluggable vector databases and model backends support enterprise use.

ARCHITECTURE OVERVIEW

User Interface Frontend (Next.js) API Routes Convex Backend

Memory Management Vector Store LLM Integration

MongoDB for metadata Vector database for embeddings Google Generative AI for responses

TECH STACK

Frontend:

Next.js 16

TypeScript

Tailwind CSS

Shadcn UI

Backend:

Express.js

Convex

MongoDB

Google Generative AI

Infrastructure:

Vercel for frontend deployment

Render for backend deployment

QUICK START

PREREQUISITES

Node.js 18+ or Bun

pnpm (preferred) or npm

Git

INSTALLATION

Clone the repository git clone cd mimic

Install dependencies pnpm install

Set up environment variables cp .env.example .env.local

Edit .env.local with your API keys

Start development servers pnpm dev

ENVIRONMENT VARIABLES

Create a .env.local file in the root directory:

NEXT_PUBLIC_CONVEX_URL=your_convex_url DATABASE_URL=your_mongodb_url JWT_SECRET=your_jwt_secret GOOGLE_API_KEY=your_google_genai_key ASSEMBLYAI_API_KEY=your_assemblyai_key

HOW IT WORKS

Create an Avatar Choose a name and avatar image. Define an initial personality using a master prompt.

Add Explicit Memories Add facts, opinions, or preferences. Each memory includes metadata such as author, trust level, and timestamp. Memories are embedded and stored in a vector database.

Start a Conversation Chat naturally. The system retrieves the most relevant, trust-weighted memories and generates responses using:

Retrieved memories

Temporary conversation context

The avatar’s master prompt

Review and Control See which memories influenced responses. Edit, delete, or re-weight memories at any time. Conversations are cleared automatically.

RUNNING THE PROJECT

DEVELOPMENT MODE

Frontend: cd apps/frontend npm run dev Runs on http://localhost:3000

Backend: cd apps/backend npm run dev Runs on http://localhost:8000

PRODUCTION MODE

Frontend: cd apps/frontend npm run build npm start

Backend: cd apps/backend npm run build npm start

PROJECT STRUCTURE

mimic apps frontend

Next.js web app

app routes

reusable UI components

shared utilities

backend

Express + Convex server

Convex functions

API routes

backend utilities

packages shared

shared types and utilities

CONTRIBUTING

Contributions are welcome.

Steps:

Create a feature branch

Commit your changes with clear messages

Push the branch

Open a pull request with a clear description

Code standards:

TypeScript for all new code

Follow existing style

Write tests for new features

Keep commits small and descriptive

DOCUMENTATION

Frontend Setup: ./apps/frontend/README.md Backend Setup: ./apps/backend/README.md API Documentation: ./apps/backend/convex/README.md Architecture Guide: ./docs/architecture.md

LINKS

Live Demo (Vercel): https://mimic-eta.vercel.app/

Live Demo (Render): https://mimic-zyb3.onrender.com

Video Demo: https://youtube.com/shorts/xD5wGwz724k?si=mCXV4I_gls4C9QjV

Support: support@mimic.ai

LICENSE

This project is licensed under the MIT License. Free to use and modify.

SHOW YOUR SUPPORT

If you find Mimic useful, consider starring the repository and sharing it.

Built with care by the Mimic team. Privacy-first. Personality-driven. Truly yours.