MakiDevelop

MakiDevelop / VirtualMe

Public

Extract a person into an AI agent — through an 8-week therapist-style interview, not a form. The open-source alternative to $4K AI clone courses.

ai-agent claude digital-twin generative-agent interview-based-agent
13
3
89% credibility
Found May 17, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

VirtualMe is an open-source tool that extracts your personality into an AI agent through an 8-week conversation instead of filling out forms. An AI interviewer asks thoughtful questions, digs deeper into your answers using a five-layer probing system, and only keeps insights that appear across multiple conversations. After two verification checkpoints, you receive eight simple text files containing your identity, values, voice patterns, skills, and boundaries—all stored locally on your machine. Your AI can then help draft messages, reply in your style, and triage requests, but always shows you drafts before anything goes out. The project costs about $60 one-time plus $5 monthly for AI service usage, compared to $3,000–$5,000 for similar commercial offerings.

How It Works

1
💡 You discover a better way

You learn that most 'build your AI clone' courses charge thousands of dollars for something you can do yourself with this free, open-source tool.

2
🤖 Your AI interviewer is ready

You connect your AI service and start a conversation that feels like talking to a thoughtful therapist who asks one question at a time.

3
🗓️ Eight weeks of deep conversation

Over 8 weeks, you have weekly sessions where the AI digs deeper into your answers, looking for patterns in how you think and what you value.

4
Your AI gets tested
✅
Passes the test

Your AI sounds authentic enough to ship

🔄
Needs more work

You do targeted follow-up conversations to fill gaps

5
📁 Your persona becomes files you own

Everything gets exported into simple text files you can read, edit, and understand—no mysterious black box.

6
✉️ Your AI helps you communicate

Now your AI assistant can draft messages in your voice, help you triage incoming requests, and always shows you drafts before anything goes out.

🎉 Your AI twin is ready

You've created a personal AI that knows how you think, speaks like you, and stays entirely under your control—no monthly subscriptions or mysterious updates.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is VirtualMe?

VirtualMe is an interview-driven pipeline that extracts a person into an AI agent through an 8-week therapist-style conversation. Instead of filling out persona forms (which people tend to fill with who they *want* to be), the system asks one question at a time and probes deeper using a five-layer decision tree: fact to pattern to principle to counter-example to triangulation. Only principles that surface across three different questions become "core truths." After 8 weeks, you get eight markdown files (SOUL, VOICE, SKILL, BOUNDARIES, etc.) plus a working agent endpoint that drafts messages in your voice. Everything stays local—no cloud sync, no human moderators. Built in Python with FastAPI, SQLite, and Claude Opus as the core LLM.

Why is it gaining traction?

The $4,000 AI clone course market sells three things: templates, API tutorials, and coaching. This project delivers the technical core for roughly $60 in API costs. Stanford research (Joon Park et al., 2024) already validated that a 2-hour interview plus LLM achieves 85% accuracy reproducing someone's General Social Survey answers. VirtualMe extends this into an 8-week production pipeline with built-in blind tests at Week 5 and Week 8 to catch overfitting before shipping. The open-source ecosystem for personal AI extraction is nearly empty—this is one of the few projects implementing the full pipeline.

Who should use this?

Technical users who want full ownership of their persona files and are willing to commit to an 8-week process. Developers comfortable running Python scripts, configuring API keys, and reviewing markdown outputs. It's less suited for non-technical users who want a plug-and-play solution. The three running plans (self, operator-assisted, cohort) give flexibility for different comfort levels with the tooling.

Verdict

VirtualMe is a serious, well-documented project with a clear academic foundation and honest limitations. The code quality is solid with type hints, async patterns, and Pydantic validation throughout. However, with 13 stars and a credibility score of 0.9%, it's early-stage software—expect rough edges and plan to read the specs before diving in. If you want a quick snapshot, Simile's MiniMe is faster. If you want depth, ownership, and the freedom to fork, VirtualMe is worth the investment.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.