Keep your reader
Readwise, BookFusion, Kindle, Apple Books, and Zotero are strong at capture, review, libraries, and citations. Highlyt starts when saved passages need to become usable thinking context.
Hi, I'm Mark.
Let me show you around.
Turn highlights intoAI-readable context.
PDF Highlighter with Knowledge Graph and MCP Server for Claude and ChatGPT
Highlyt lets you color-code ideas by meaning, link them across books and papers, and connect Claude or ChatGPT directly via MCP. Your highlights stop being dead text and become source-backed memory.
highlyt.app
Highlights
Graph
The problem
You read constantly. You retain almost nothing.
That paper you read last Tuesday? The one with the perfect framework? You highlighted it. You know you did. But now it's buried in a folder somewhere, and you're re-reading the whole thing.
"I highlighted this exact thing last month..."
You remember the insight. You remember the yellow highlight. Maybe it was in Atomic Habits, maybe that research paper on habits. You open 6 PDFs, scroll through hundreds of pages, and pray. Most of the time, you give up and rewrite it from memory.
Your AI treats every highlight the same
You paste 3 pages of book highlights into Claude. It treats everything equally. It doesn't know that paragraph was a key framework, this one was a question you had, and that one was supporting evidence. You did that thinking while reading. But none of it survived the copy-paste.
Frameworks from 5 books, zero connections
At 11 PM you realize the decision-making framework in Thinking Fast and Slow perfectly explains the bias in your research data. It's the kind of insight that could anchor your whole argument. But there's nowhere to save that connection. By morning, it's a vague feeling that two things were related.
What Highlyt is
A synthesis layer for highlights you already know are valuable.
Highlyt is not trying to be every reader, import inbox, or daily review system. It is for the moment after capture, when hundreds of saved passages need meaning, relationships, and source-backed reuse.
Connect what mattered
Semantic colors preserve why you highlighted something. Labeled links preserve how ideas relate: supports, contradicts, expands, questions, or related.
Give AI the structure
Claude and ChatGPT can read highlights, color meanings, source metadata, notes, graph edges, and cluster context through MCP instead of a flat pasted dump.
What if every highlight from every book and paper was instantly useful?
Searchable. Color-coded. Connected across documents. And ready to feed into your AI the moment you need it.
Colors that carry your reasoning
Yellow doesn't just mean "important." You decide what it means. Yellow = key concept, blue = question, green = framework. Change it per book or paper. When Claude reads your highlights via MCP, it knows that yellow passage is a mental model, not just random text. Your thinking survives the handoff.
Color definitions
Key concept12
Question8
Evidence15
Methodology6
See how your ideas actually connect
That framework in Chapter 3 of Thinking Fast and Slow explains the result on page 90 of your research paper. Now you can literally draw that line. Link any highlight to any other, across chapters, books, and documents. Watch an interactive knowledge graph emerge from your reading.
5 nodes, 6 connectionsView full graph →
Find any highlight in under a second
That quote from Atomic Habits three months ago? Type two words and it's there. Search across every book, every paper, every highlight, every note you've ever made. Filter by color, by chapter, by document. No more scrolling through 40 books to find one framework.
transformer attention mechanism
Attention Is All You Needp.3
...multi-head **attention** allows the model...BERT: Pre-trainingp.7
...bidirectional **transformer** architecture...GPT-4 Technical Reportp.12
...scaled dot-product **attention mechanism**...Connect Claude and ChatGPT via MCP
Connect Claude directly to your highlights via MCP. It sees your color-coded reasoning, your linked ideas, your notes. Not a wall of copy-pasted text. Export as structured JSON or Markdown when you need it elsewhere. Your AI finally gets the context you spent hours building.
MCP Server
Active$ claude mcp connect highlyt
Connecting to highlight server...
✓ 47 highlights from 5 documents ready
✓ Knowledge graph with 12 connections
Connecting to highlight server...
✓ 47 highlights from 5 documents ready
✓ Knowledge graph with 12 connections
From PDF to knowledge graph in 5 minutes
No setup, no plugins, no learning curve. Just start reading and let your knowledge compound.
01
Upload any PDF
Drag in any book, research paper, textbook, or article. Text is extracted instantly and every word becomes searchable from the moment it uploads.
P
research-paper.pdfP
thesis-ch3.pdf02
Highlight with meaning
Select text and annotate it with a color that carries your reasoning. Yellow = key concept, blue = question, green = evidence. Each document can have its own color meanings. Your highlights now carry intent, not just color.
Attention is a mechanism for computing...Key concept
Does this generalize to longer sequences?Question
BLEU score improved by 2.0 over existing...Evidence
03
Link ideas into a knowledge graph
Draw connections between any two highlights, across chapters, books, and papers. Watch an interactive graph emerge that reveals the structure of your thinking. The mental models you noticed while reading are now permanent and visible.
Key concept Question Evidence
04
Feed your AI the full picture
Plug Claude in directly via MCP and it sees everything: your highlights, color meanings, linked frameworks, notes. Or export as JSON/Markdown. Your AI finally understands what you read the way you understood it.
AI
Claude via MCPConnected{ "highlights": [
{ "text": "...", "color": "key_concept" },
{ "text": "...", "color": "question" }
] }
{ "text": "...", "color": "key_concept" },
{ "text": "...", "color": "question" }
] }
Everything you need for smarter research
Semantic colors
Full-text search
Knowledge graph
MCP for LLMs
Structured export
Shelves & organization
Private & secure
Coming soon
Chrome extension
Pricing
Less than your morning coffee
No credit card required. Cancel anytime.
Early adopter pricing. Lock in this rate before we raise prices.
Free Trial
7 days, no credit card
$0to start
- 3 PDF/EPUB documents
- Unlimited web clips and YouTube
- Color-coded highlights
- Knowledge graph
- MCP server access
- Full-text search
Most Popular
Pro
Everything, no limits
$5.99$3.25/month
Billed $39.00/year
- Unlimited PDFs and EPUBs
- Unlimited web clips and YouTube
- Up to 100 MB per file
- Unlimited highlights and links
- Knowledge graph
- MCP server for Claude and ChatGPT
- Full-text search across everything
- JSON and Markdown export
- Notes, shelves, and organization
- Priority support
Web clips and YouTube saves are unlimited on all plans. Document limits apply to PDFs and EPUBs only.
Your next book deserves better than a yellow highlighter
Every book and paper you read builds your knowledge. Highlyt makes sure none of it gets lost. Start your free trial and turn scattered highlights into a connected, searchable, AI-ready knowledge base.
Knowledge graph that grows with youClaude and ChatGPT read your highlightsEvery color carries your reasoning
Highlyt
You highlight. You forget. Highlyt connects the dots.
FAQ
Frequently asked questions
What is Highlyt?+
Highlyt is a synthesis layer for serious reading. Upload PDFs, EPUBs, web articles, or supported Kindle highlights; mark passages with semantic color meanings; link ideas across documents; and turn those annotations into source-backed context for Claude or ChatGPT.
How is Highlyt different from Readwise or Zotero?+
Readwise is stronger for broad highlight capture and daily review. Zotero is stronger for citations and reference management. Highlyt is built for what happens after capture: semantic color meanings, labeled links between highlights, graph structure, source citations, and AI-readable context.
Can I use Highlyt for books, not just research papers?+
Yes. Highlyt works with any PDF, whether it is a business book, self-help book, textbook, or research paper. Book readers use it to map mental models and frameworks across multiple books, search highlights from their entire reading history, and feed book insights to AI for synthesis and application.
What is an MCP server and how does it work with Highlyt?+
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools. Highlyt's MCP server gives Claude or ChatGPT access to your highlights, color meanings, linked ideas, notes, and source metadata, so AI can reason over structured reading context instead of a flat export.
How do I create a knowledge graph from my highlights?+
Link any two highlights by selecting them and choosing "create link." You can connect ideas across pages, chapters, and different documents. Label relationships like "supports," "contradicts," or "extends." An interactive visual graph emerges showing how your ideas connect. The patterns you noticed while reading become permanent and visible.
What does color-coded semantic highlighting mean?+
Standard highlighting treats all colors the same. Semantic highlighting assigns meaning to each color. For example: yellow for key concepts, blue for open questions, green for supporting evidence, pink for methodology. You define meanings per document. When exported or read by AI, the colors communicate your reasoning, not just that something was marked.
How much does Highlyt cost?+
The Pro plan costs $3.25 per month billed annually ($39 per year) or $4.99 per month billed monthly. This includes unlimited documents (up to 100 MB each), unlimited highlights, knowledge graph, MCP server access, full-text search, and JSON/Markdown export. A 7-day free trial with 3 documents (up to 25 MB each) is available with no credit card required.
Is my data private and secure?+
Yes. Your PDFs and highlights are stored securely using Supabase with Row Level Security (RLS). Only your authenticated account can access your data. No one else, including Highlyt, can read your documents or highlights. Your data is never shared with third parties or used for AI training.
Can I export my highlights to use with other AI tools?+
Yes. Export your highlights as structured JSON or Markdown, complete with color meanings, page numbers, notes, and linked ideas. This works with any AI tool, note-taking app, or workflow. For Claude and ChatGPT specifically, the MCP server provides a direct connection without needing manual export.
What is a knowledge graph in a PDF reader?+
A knowledge graph is an interactive visual map of connections between your highlights. Link any two highlights across pages or documents and label the relationship. The graph reveals patterns in your reading that linear notes cannot capture. It is especially useful for mapping mental models across multiple books or connecting evidence across research papers.
Have more questions? Email us at [email protected]