An AI simulation tool capable of building “digital worlds” to predict the future has caught the attention of Sequoia China, Hillhouse Capital, and other top-tier investors. The team behind it? A recent graduate who built the core technology in just 10 days.
MiroFish, an open-source AI simulation engine, has raised $4.2 million (30 million RMB) in angel funding, transforming a graduation project into one of the most discussed AI initiatives in China’s developer community.
What Is MiroFish?
MiroFish is an AI simulation platform that creates digital twins of complex systems — social networks, financial markets, supply chains, even entire cities. The goal: predict how these systems will evolve by running millions of simulated scenarios.
Unlike traditional simulation tools that require painstaking manual modeling, MiroFish uses large language models to automatically generate and calibrate digital agents that behave like real people, companies, or organizations.
Imagine running 10,000 versions of a stock market, each with AI agents making different decisions, to see which patterns emerge most frequently. That’s the power of MiroFish.
The 10-Day Origin Story
According to the project’s GitHub repository, the core simulation engine was built in just 10 days as a university graduation project. The creator, operating under the handle 666ghj, designed the system to explore how AI agents interact within complex social systems.
- Days 1-3: Core simulation framework and agent architecture
- Days 4-6: LLM integration for realistic agent behavior
- Days 7-10: Visualization tools and scenario testing
What started as an academic experiment quickly gained thousands of GitHub stars and attracted attention from researchers and investors alike.
Why Investors Are Betting Big
The participation of Sequoia China and Hillhouse Capital signals strong confidence in simulation-as-a-service as a viable business model. The technology addresses growing demand across multiple industries:
- Finance: Stress-test portfolios against thousands of market scenarios
- Supply Chain: Model disruptions and optimize logistics
- Urban Planning: Simulate traffic, population growth, and resource allocation
- Marketing: Predict viral content and consumer behavior
MiroFish’s open-source nature has accelerated adoption. Developers can inspect the code, contribute improvements, and build custom applications on top of the platform.
How It Works
MiroFish combines several AI technologies into a unified simulation engine:
- LLM-powered Agents: AI agents that can reason, plan, and make decisions like real stakeholders
- Multi-Agent Orchestration: Framework managing thousands of interacting agents
- Environment Modeling: Tools for defining rules, constraints, and initial conditions
- Result Aggregation: Analytics to identify patterns across simulation runs
The system can be configured via Python APIs or a web interface, making it accessible to both developers and non-technical users.
What’s Next
The new funding will expand the engineering team, improve the core simulation engine, and develop industry-specific solutions. The team is also exploring partnerships with research institutions to validate the platform’s predictive accuracy.
For now, MiroFish remains open-source under the MIT license, with a growing community of contributors pushing the boundaries of what AI simulation can achieve.
When we started, we just wanted to see if AI agents could simulate realistic social dynamics. We never expected this level of investment.
Getting Started
Developers can explore MiroFish on GitHub. The repository includes documentation, example scenarios, and a quick-start guide for building your first simulation.
The project is actively maintained, with regular updates adding features and improvements based on community feedback.
