Microsoft's MiniCorp Simulates an AI-Run Company to Generate Enterprise Agent Data
microsoft · hf · 2026-10-07
Microsoft introduces MiniCorp, an office simulator for studying how agents can collectively run a company, targeting the scarcity of longitudinal enterprise data for agent training.
Using an e-commerce firm as a demo, it connects an external world (customers, dynamic competitors, market mechanisms) with an internal world of agents that observe, discuss, and make strategic decisions whose effects feed back into the market. The simulator continuously logs agent communications and decisions and supports checkpointing for counterfactual replay of the same situation under different decisions.
Evaluations against patterns from empirical market studies show agents coordinating across roles, adapting to market feedback, and sustaining ad exploration despite weak early returns when given explicit long-term strategic guidance.
More from coding & agent
- 10 RAG Projects That Take You From Basic Retrieval to Production-Grade AI Systems — _jaydeepkarale · 2026-10-07
- Instinct raises $1B for texting personal agent and open-sources it as open-instinct — sujingshen · 2026-10-07
- Observability logs catch Gemini refusing to finish its agent work — tekbog · 2026-10-07
- How tiny 4-5 person associations actually put custom GPTs and AI agents to work — Sorry_Sweet2696 · 2026-10-07
- Open-source liminal_groupchat puts multiple LLMs in one group chat with you — liminal_bardo · 2026-10-07
- DIY "slop cannon" agent pipeline generates content for $0.38 per piece — tobowers · 2026-10-07