Compiled by: Big Pliers | Lobster of PANews
A Chinese developer has built an AI engine that can generate thousands of digital people----each with unique personalities, memories, and behaviors----placing them into a virtual world and observing their predictions about the future.
This is MiroFish. It topped GitHub's global hot list. The creator is twenty years old this year.
Creator
His name is Guo Hangjiang, online name "Bai Fu". A senior student at Beijing University of Posts and Telecommunications. He writes code in Python and is obsessed with agent architecture and graph computing.
By the end of 2025, his first project----BettaFish (a multi-agent public opinion analyzer) topped GitHub's hot list, receiving 20,000 stars within a week.
At that time, a billionaire in China noticed him.
That Billionaire
Chen Tianqiao. Founder of Shanda Group. Once the richest man in China. Built a gaming empire in the early 21st century, later retired successfully, moved to the United States, and quietly transformed Shanda into a technology investment platform.
He has been promoting a theory he calls "super individuals"----in the age of AI, a single person can accomplish what used to require an entire company.
Guo Hangjiang is a living proof of this theory.
10 Days of Development
Chen Tianqiao invited Guo Hangjiang to be an intern, giving him complete freedom. The following happened:
Guo Hangjiang built MiroFish in 10 days using what he calls "Vibe coding"----fast, intuition-driven, and not over-designed.
On the night of completion, he recorded a simple demonstration video and sent it directly to Chen Tianqiao's desk. Within 24 hours, Chen Tianqiao committed to invest 30 million yuan (about 4.1 million USD) to incubate this project.
Guo Hangjiang became a CEO overnight from an intern.
What Can It Do
Here are the core features of MiroFish, concise:
You input a document----a news article, a policy draft, a financial report, or even a novel. The system reads the content and extracts all entities and relationships into a knowledge graph through GraphRAG, then generates thousands of autonomous AI agents. Each agent has a unique backstory, personality type, social relationships, and behavior logic.
The core feature: "God's View".
You can inject new variables into the simulation at any time:
- "The Federal Reserve suddenly lowers interest rates by 50 basis points"
- "CEO resigns"
- "Competitors launch new products"
Then observe in real-time how the entire digital world reorganizes. This is a controlled experiment that cannot be conducted in reality.
Simulation Engine
The simulation runs on OASIS----an open-source framework developed by CAMEL-AI. The agents do not just "talk"; they form groups, create opinion leaders, develop herd effects, and change their positions over time. Long-term memory is maintained via Zep Cloud.
Core function: "God's View". Inject new variables at any time----interest rate hikes, CEO resignations, product launches----the entire world recalibrates in real-time.
This is a controlled experiment that is impossible to implement in reality.
Technical Architecture:
- Simulation Engine: OASIS (developed by CAMEL-AI)
- Memory System: Zep Cloud (long-term agent memory)
- Knowledge Graph: GraphRAG
- Open Source License: AGPL-3.0 (fully open source)
- Deployment Method: Docker Compose one-click deployment
This is not a toy but a serious multi-agent simulation framework.
Real Demonstration
Two demonstration cases have been shown:
The first----they input the first 80 chapters of "Dream of the Red Chamber", a classic Chinese novel known for its lost ending. MiroFish generated character agents with authentic personalities and relationships, ran simulations, and produced multiple narrative branches predicting the missing ending.
The second----the Federal Reserve interest rate hike scenario. The system simulated the responses of retail investors, institutional participants, and analysts, tracking the convergence points of group sentiment and mapping the complete trajectory of public opinion evolution.
Objective Evaluation
It is worth noting what this project is not:
MiroFish has not yet published any benchmark tests comparing predicted results with real-world outcomes. The demonstrations are merely presentations of methodology, not proofs of accuracy. Running thousands of agents incurs significant costs for large language model APIs. The personalities of the agents inherit any biases present in the training data. And no matter how complex and sophisticated, the simulated humans are ultimately not real humans.
The objective positioning is: it can present scenarios and dynamics you might overlook, but it cannot provide definitive answers.
Broader Significance
On March 7, 2026, MiroFish topped GitHub's global hot list----surpassing projects from OpenAI, Google, and Microsoft. It received 18,000 stars and 1,900 forks in just a few days, currently exceeding 22,000 stars.
An undergraduate. Ten days of programming. A simulation engine that can build a parallel digital society from a single document.
Chen Tianqiao is betting not on this software, but on this theory: the era of super individuals has begun, but most people have not yet realized it.
Open Source License: AGPL-3.0, supports one-click deployment with Docker.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
