Hey Product Hunt! I’m Cristian (Eno), founder of Omniamus and maker of Oxyo (https://oxyo.si).

If you build every day with Claude Code, Cursor, Codex, or Gemini, you know the two biggest productivity killers: every new chat starts from zero (burning tens of thousands of tokens re-reading context), and complex math or architecture tasks spiral into expensive trial-and-error loops.

I built Oxyo to give your existing AI coding tools a persistent intelligence layer over Remote MCP (https://oxyo.si/mcp) in 30 seconds:

45–72% lower token & quota burn across real multi-session engineering workflows—getting 2–3× more coding capacity out of the subscription you already pay for.
Zero session amnesia: active project rules and verified solutions carry across chats without pasting massive logs.
Measured on hardware token counters: in our head-to-head Homo Vitruvianus test on Gemini 4 Carbon (High), connecting Oxyo cut output & reasoning tokens by 48.5%, reduced thinking overhead by 82.8%, and finished 2.66× faster while delivering higher geometric & 3D visual fidelity.

You can inspect the live 3D demos and full telemetry at https://oxyo.si/benchmarks, or start free (50 MCP calls included, no infrastructure to manage) at https://oxyo.si.

Would love your feedback and tough benchmark prompts in the comments!

Hey Product Hunt! I’m Cristian (Eno), founder of Omniamus and maker of Oxyo (https://oxyo.si).

I built Oxyo to give your existing AI coding tools a persistent intelligence layer over Remote MCP (https://oxyo.si/mcp) in 30 seconds:

45–72% lower token & quota burn across real multi-session engineering workflows—getting 2–3× more coding capacity out of the subscription you already pay for.
Zero session amnesia: active project rules and verified solutions carry across chats without pasting massive logs.
Measured on hardware token counters: in our head-to-head Homo Vitruvianus test on Gemini 4 Carbon (High), connecting Oxyo cut output & reasoning tokens by 48.5%, reduced thinking overhead by 82.8%, and finished 2.66× faster while delivering higher geometric & 3D visual fidelity.

You can inspect the live 3D demos and full telemetry at https://oxyo.si/benchmarks, or start free (50 MCP calls included, no infrastructure to manage) at https://oxyo.si.

Would love your feedback and tough benchmark prompts in the comments!

Oxyo

Cut AI coding token burn by up to 72% via Remote MCP

Cut AI coding token burn by up to 72% via Remote MCP