What is Caimeo?

A Local-First Framework for Deterministic AI Simulations

Caimeo is an open-source framework for creating autonomous AI agents that can run in **verifiable, deterministic** environments—either **fully offline** or integrated with on-chain/TEE systems for advanced proofs. Our core mission is to enable developers to deploy, analyze, and test AI logic without centralized dependencies, preserving privacy and guaranteeing reproducible results.

                        Core Capabilities
                        **Offline** agent simulation with deterministic random seeds
Flexible environment definitions (grid- or scenario-based)
Integrated memory and modular agent logic (LLMs, state machines, etc.)
Verifiable proofs of simulation results (on-chain or via TEE attestations)
Optional bridging to blockchain for advanced use cases

                    

                        Key Benefits
                        **No middlemen**: purely local-first or offline execution possible
**Deterministic**: identical inputs yield identical outputs every time
**Modular**: drop-in AI modules, environment expansions, and more
**Secure**: uses Trusted Execution Environments (TEEs) or ZK to guarantee fairness
**Extendable**: easy to integrate with external systems, or embed in gaming, robotics, research, etc.

                    

Technical Foundation

Built on open standards for maximum extensibility

Deterministic Simulation Engine

The heart of Caimeo is a tick-based simulation loop. Each step processes every active agent (AI or environment actor) according to a pre-seeded random number generator. This ensures any offline or on-chain re-run with the same seed yields identical results.

Environment Plugins & Expansions

Caimeo uses a plugin-based architecture for environment logic. You can equip expansions for hazards, dynamic conditions (e.g., fires, storms), or entirely new rule sets. This modular approach keeps the core engine simple while allowing advanced scenario-building.

TEE & Zero-Knowledge Support

For trustless or privacy-sensitive use cases, Caimeo integrates with:

**TEE** solutions (e.g., Phala Network) for confidential simulation runs
**ZK proofs** (planned) to attest to results without revealing agent logic

Multi-Model LLM Support

The framework accommodates large language models or simpler AI modules. You can plug in any GPT-like or open-source LLM (Llama, OpenAI, Anthropic, etc.) to power agent decision-making. Caimeo leaves the choice of AI backend entirely up to you.

How Caimeo Works

Local or remote—always deterministic

Offline or Local-First Mode

You can run the Caimeo engine entirely offline, specifying an environment definition (2D/3D layout, hazards, event triggers, etc.) and a set of agent logic modules. Caimeo uses a **seed** to ensure everything is perfectly reproducible. Developers can iterate quickly without any chain interactions.

TEE-Verified Mode

When outcomes require trustlessness or confidentiality, Caimeo can run inside a **Trusted Execution Environment** (e.g., Phala). The TEE keeps private agent logic secure while producing a verifiable on-chain attestation of the final results. This eliminates tampering, ensures fairness, and keeps advanced AI code hidden from adversaries.

On-Chain Checks or Full On-Chain Execution

Caimeo can also function in a partial on-chain environment: a contract can read the TEE output or a ZK proof to confirm the legitimacy of a run. In simpler cases, the entire environment step logic can be ported on-chain (though that may be expensive). The key is that **Caimeo is flexible**—pick your trust model and complexity level.

Example Use Cases

Where Caimeo can make a difference

Crisis Simulations

Test emergency response strategies offline. Tweak environment plugins (fire spread, intruder logic, resource constraints) and run thousands of ticks to find best outcomes. Move verified results on-chain if you need public trust.

Game Development & AI NPCs

Build advanced AI for game NPCs that can run or “train” offline. Import them into your actual game with guaranteed performance based on prior verifiable test runs. Perfect for emergent or unpredictable NPC behavior.

Market or Economic Simulations

Simulate supply-demand, trading, or macro-economic shifts with AI-driven participants. Keep the entire simulation local or, if needed, post high-level proofs on-chain to demonstrate fairness and consistency of results.

Research & Decision Support

Environmental modeling, robotics route planning, or any scenario requiring repeated simulations with guaranteed reproducibility. Use Caimeo to automate experimentation, gather data, and refine your AI strategies.

Relation to NFTs & "Caimeo Agents"

**Caimeo** is primarily a self-contained framework for AI simulation. It can optionally connect to NFT-based expansions when you want agent identities, environment ownership, or reward distribution to live on-chain. An example of this is the Caimeo Agents (and "Agent Games") project, where:

Agents and Environments can be tokenized (NFT or SBT) to track ownership or modifications
On-chain "proof of outcome" is used to distribute rewards or new abilities

However, this NFT aspect is **not required** to use Caimeo for local or private simulations. It is simply one way to support open, decentralized communities in verifying agent performance. To learn more about how Caimeo is extended into a full NFT-based ecosystem, check out Caimeo Agents.

Get Started

Caimeo is open source and welcomes collaboration. If you want to contribute or use Caimeo for your own AI simulations, keep an eye on our GitHub repository (coming soon). If you’re curious about the NFT-based expansions or want to back the project, you can explore Caimeo Agents as well.