Why I built Alyph.

I spend roughly $2,000 a month using LLMs to build software. And I was constantly losing context in linear AI chat.

In standard linear chats, every prompt stacks into one long history log. When I was building out features, Feature B would inevitably get polluted by Feature A's logic. If I explored a dead end ten prompts deep, my entire workspace was permanently cluttered with useless context.

I needed a way to anchor my codebase to a root node and branch off isolated sub-threads that never leaked into each other. I needed to test different approaches from the same model, and I wanted to send the exact same prompt to ChatGPT and Gemini simultaneously to compare outputs side-by-side.

Standard chat interfaces couldn't do this. So I built Alyph. It is a visual LLM canvas where you can execute parallel features, try different models instantly, and prune deep context with a single click.