Under the hood, without the jargon. Everything Cyton does is modelled on how your brain actually stores, connects, and recalls information.
Neuroscientists have identified five distinct memory systems in the brain, each serving a different purpose. Cyton uses the same model — every note you write is automatically classified into one of these five tiers based on how you interact with it.
You never manually sort a note. Cyton watches how recently you wrote it, how often you revisit it, how many other ideas it connects to, and how it appears in your conversations. From these signals, it decides where each note belongs — and updates that classification as your habits change.
Notes you just wrote or recently captured. Like a first impression — your brain holds it briefly while deciding whether it matters. Sensory nodes decay quickly if not revisited.
Ideas you're actively thinking about — notes you've opened, edited, or mentioned in conversation in the last few days. These are the concepts currently loaded into your mental workspace.
Time-stamped experiences and dated entries. Notes tied to a specific moment — a meeting summary, a journal entry, a decision you made. Episodic memory anchors ideas to when they happened.
Core knowledge you've reinforced enough times that it's become stable. Concepts you've revisited consistently over weeks or months. These are your 'known' ideas — the foundation your thinking builds on.
Step-by-step processes, how-to guides, and repeatable workflows. Knowledge that's been abstracted into a pattern — not just what you know, but how you do things.
How tiers shift over time. A note you wrote yesterday starts in Sensory Memory. Revisit it three times this week and it moves into Working Memory. Review it consistently over months and it solidifies into Semantic Memory — your established knowledge base. Stop engaging with it entirely, and it gradually fades back toward the edge of your graph. Cyton updates tier assignments continuously — not as a one-time import.
Every edge in Cyton's knowledge graph is earned through use — not inferred from keywords or set up manually. The source of truth is your actual conversations with Claude Code.
When you work with Claude Code through Cyton's MCP connection, Cyton monitors which topics and concepts come up together in your sessions. If you discuss startup strategy and LLM agents repeatedly in the same sessions, the edge between those two nodes gets stronger. If you stop discussing a topic entirely, its connections gradually weaken — the same way a relationship you never maintain fades.
Connect your Obsidian vault to Cyton once, and it maintains a live mirror of every note you write. This isn't a one-time import — it's a continuous background process. When you edit a note in Obsidian, Cyton detects the change and updates its graph accordingly, usually within seconds.
You don't need to change how you use Obsidian. Write notes exactly as you always have. Cyton works in the background, extracting the structure and meaning from what you've written — without touching your files.
Memory researchers discovered that forgetting follows a predictable curve — and that reviewing something just before you'd otherwise forget it is the most efficient way to make it stick. The longer you hold onto an idea before forgetting it, the less often you need to review it going forward.
Cyton tracks your review history for every note and schedules the next review based on how stable that memory has become. A note you've reviewed ten times needs far less attention than one you've seen twice. The queue only shows you what actually needs reviewing — not everything.
Pro: REM pattern completion. During nightly consolidation, Cyton's Pro tier runs a deeper analysis of your graph — looking for patterns, surfacing connections between ideas you haven't explicitly linked, and generating review suggestions for the next day. Think of it as the equivalent of what your brain does while you sleep.
Cyton connects to your AI tools via MCP — the Model Context Protocol, developed by Anthropic. This lets your AI assistant (Claude Code, or any other MCP-compatible model) read from and write to your knowledge graph as it works with you.
Cyton's cloud uses Claude Haiku to automatically enrich your graph in the background — extracting concepts from new notes, strengthening relationships, and flagging nodes for review. This runs on Cyton's infrastructure. You get the benefit without needing your own API key.
Connect your own AI model — Claude Sonnet, GPT-4o, Gemini, or any other MCP-compatible model — and Cyton routes all enrichment through it. Everything runs on your machine. Your notes are never sent to Cyton's servers. You choose the model. You own the data.
Download Cyton free. The full knowledge graph, five memory tiers, and Obsidian sync — no account required to start.