I’m building TinyAgentOS or taOS, a self-hosted collaborative AI agent and project workflow focused OS that runs across whatever hardware you have: old laptops, SBCs, gaming rigs, or small clusters. It focuses on auto clustering, resource aware scheduling, and running local AI agents without depending on expensive cloud infrastructure.

Alongside that, I’m developing taOSmd, a framework agnostic AI memory system designed to give agents longterm, high recall memory without any cloud dependency. It’s a standalone library that powers TinyAgentOS’s agent memory, achieving strong results on long context benchmarks like LongMemEval-S while staying simple to integrate into other stacks.

I actively test this ecosystem across real hardware: ARM boards, consumer GPUs, and hybrid home lab style clusters. That brings ongoing costs for boards, hardware, networking, and power, plus the time to reproduce bugs, benchmark memory behaviour, and keep up with rapidly changing model/tooling ecosystems.

Your sponsorship directly helps with:
• Expanding and maintaining diverse test hardware for TinyAgentOS and taOSmd
• Covering server, bandwidth, and power costs for always-on clusters and benchmarks
• Investing time into docs, examples, and integrations so taosmd works well beyond TinyAgentOS
• Keeping both projects fully open-source and sustainable over the long term

If you care about local AI, edge compute, or giving agents durable, high quality memory on your own hardware, your support helps me move faster, test on more setups, and make this ecosystem as solid and accessible as possible.