Trenin Labs is an independent lab building the practical infrastructure behind production AI — from the skills your agents use, to the hardware they run on. We ship the boring, load-bearing pieces that the rest of the stack assumes.
Two products tackling two real gaps in the AI stack: the skills agents use, and the hardware they need to run.
A self-improving skill library for AI agents — over MCP
An MCP server that delivers state-of-the-art skills to your agents, captures their feedback in production, and rolls improvements back to every product using them. One skill, every environment, always getting sharper.
Visit AgentArmory.AiHardware recommendations for running LLMs
Pick the right rig for the model you want to run. Compare builds across use cases, memory budgets, and specific open-source LLMs — and get a setup that actually fits your workload before you spend on GPUs.
Visit LLMRequirements.comTrenin Labs is a Netherlands-based company building practical infrastructure for the next generation of AI systems. We work at the intersection of applied research and product engineering — shipping the missing pieces between a clever prompt and a production-grade agent.
We’re small, independent, and focused. If you’re building something serious with LLMs or autonomous agents, we’d love to hear about it.