Today, I’m excited to unveil ForbocAI.
The problem with current Game AI is that it’s static—decision trees that always lead to the same place. The problem with extensive GenAI integration is that it’s chaotic—hallucinations that break game immersion and mechanics.
We are building the bridge: Neuro-Symbolic Agents.
At ForbocAI, we use deterministic systems (ECS) to handle the rules of physics and combat, while layering local-first Small Language Models (SLMs) to handle high-level reasoning, dialogue, and memory.
This isn’t just a chatbot in a game. It’s an agent that:
<ol>
<li>Observes the world (via vector embedding streams).</li>
<li>Reasons about its goals (using a local RAG buffer).</li>
<li>Acts deterministically (executing verified game commands).</li>
</ol>
And soon, you will own them. We are working on a decentralized framing that lets you train an NPC and take its “Soul” (logic + memory) to other worlds.
We are opening our Seed Round for strategic partners who see the future of Autonomous Worlds.
Check out our roadmap: <a href="https://forbocai.github.io" target="_blank" rel="noopener nofollow">https://forbocai.github.io</a>

Sean  Dinwiddie

ForbocAI introduces neuro-symbolic agents for game AI that combine deterministic Entity Component Systems (ECS) for physics and combat with local Small Language Models (SLMs) for reasoning, dialogue, and memory. These agents observe the game world through vector embeddings, reason using local RAG buffers, and execute verified game commands. The platform aims to enable decentralized NPC training where players can transfer an NPC's logic and memory across different game worlds.

We’ve been quiet, but we’ve been building.