Bridge the gap between reading and AI

It started with a paper on cognitive biases. I had highlighted a framework in Chapter 4 of Thinking, Fast and Slow that perfectly explained the results in a research paper I was reading. I knew the connection was there. I could feel it. But there was nowhere to record it. I had highlights in one PDF, highlights in another, and no way to link them.

By morning, the connection was a vague feeling that "two things were related." That was the moment I decided to build something.

Not another PDF reader. Not another note-taking app. A tool that treats highlights as first-class objects with meaning, relationships, and context.

Highlyt lets you color-code your highlights with semantic meaning. Yellow doesn't just mean "important." You decide what it means: key concept, question, evidence, methodology. Then you link highlights across documents into a knowledge graph. The pattern you noticed at 11 PM between two books becomes a permanent, visible connection.

When you connect Claude or ChatGPT via MCP, your AI sees the reasoning behind your highlights, not just the text. It knows this passage is a mental model, that one is a question, and these two are connected. Your reading becomes a conversation partner.

Today, Highlyt is used by students preparing for exams across multiple textbooks, researchers running literature reviews across dozens of papers, and non-fiction readers mapping frameworks across their entire reading history. The common thread: they all highlight seriously and were frustrated that their highlights went nowhere.