All Work

Sole Architect

Simulation Lab

Two social simulations, same input, one synthesis, macro population dynamics and individual agent cognition in parallel.

Context

Focus groups are slow. A/B testing requires live traffic. Neither works before you've spent money.

Strategy

Dual simulation: MiroFish for breadth (population dynamics), AgentSociety/Tsinghua for depth (individual agents with emotions, Maslow needs, LLM cognition).

Solution

Convergent findings (high confidence), divergent findings (where the insight lives), confidence levels.

The Architecture

MiroFish models populations as interacting agents with influence networks. AgentSociety gives each agent biography, needs state, decision-making architecture.

The Synthesis Layer

Convergent findings, divergent findings, confidence levels. Divergence is diagnostic, population vs individual motivations pulling different directions.

Use Cases

Market entry, public opinion, organizational change.

2
Simulation engines
||
Parallel execution
%
Confidence-scored
MiroFish AgentSociety Opinion Dynamics Behavioral Prediction LLM Agents