this is very much beta software. for installation instructions and up-to-the-update information, visit enough.support.
enough is an extensible personal language system with local models and openrouter support for planning, reviewing, writing, and translating text(s). it's optimized for so-called non-technical users, and built to be useful and usable for decades.
it's built upon GGML, llama.cpp, and whisper.cpp by Georgi Gerganov, the superb GGUF-work of Unsloth, FastAPI by Sebastián Ramírez, HTMX by Carson Gross, UV by Astral, and many other wonderful free and open source software projects.
enough is designed for people who want to enjoy the many benefits of LLMs without giving up control of their data or being bound to hosted or otherwise restrictive inference providers. it's designed to be useful for private local work right now, but to remain available and be updated to take advantage of whatever new models and capabilities emerge over the next 5, 10, or 50 years (provided a community continues to want and support it).
in its current form, enough is best suited for text-based knowledge work: creative writing, journaling, synthesis, exegesis, research, etc.
however, the only actual limitations are what the supported models can do on your machine ... and your imagination.
this is because enough imposes very few paradigms on your wordflows -- in a world of exponential possibility, most people have as many things they want to do as ways they prefer to do them.
enough's core design principles make it easy and fun for anyone to build out their own "knowledge operating system" with global- and project-level agent paradigms, skills, roles, and knowledge stores.
enough.support is also a forum where users can get help with anything they want to build into their setups, provide assistance to others, and/or just show off the "thought-enhancing prosthetics" they've cooked up.
enough is built and tested on MacOS & Apple Silicon. Linux support is planned; Windows support is totally feasible. the feature set may also eventually expand beyond this single-viewport agent harness.
for now, the idea is that this might be enough for quite a while.
© 2026 Graham Smith — released under the Apache License 2.0