Game-theoretic AI that computes provably optimal strategies.
Most strategic AI gives advice that is probably good. Mieza computes advice that is provably optimal — Nash equilibria grounded in your actual decision structure, not heuristics or pattern-matching.
The platform quantifies exploitability: the amount an adversary can extract from you beyond what they would get if you played optimally. When stakes are high and decisions repeat, the ROI on closing that gap compounds.
- Equilibrium solvers — CFR for sequential games, support enumeration for simultaneous-move games
- Natural-language translation — solver output converted into actionable recommendations via AI strategist agents
- Integration layer — MCP server for LLM/AI assistant connections, REST API, SDKs
Pricing strategy, negotiation, compensation design, competitive positioning, resource allocation, mechanism design, adversarial security.
| Repository | Description |
|---|---|
| mieza-clj-sdk | Clojure SDK — game protocols and domain entity schemas |
| mieza-api-clj | Clojure API client (Martian-based, dynamic from OpenAPI) |
| mieza-api-python | Python API client (auto-generated from OpenAPI) |
| mieza-game-compensation | Compensation negotiation game — effort, retention, vesting |
| mieza-game-pricing | Competitive pricing game — heterogeneous buyers, menu design |
- mieza.ai — Platform
- joshuacol.es — Founder's writing on game theory, compression, and equilibrium computation