Research
Turn a folder of papers into a citation graph your AI can actually reason over.
Give Every AIA Knowledge GraphThat Remembers.
Give Every AI
A Knowledge Graph
That Remembers.
Nitanics turns any folder of code, docs, research or images into a persistent, queryable graph — plugged into Claude, Gemini, Codex, Cursor, Copilot or any AI over MCP.
Your AIShouldn'tRestart From Scratch Every Session
Re-reading the same files
Every session your AI scans the repo from scratch — burning tokens on context you already paid for.
Research stranded in silos
Papers, PDFs, repos and notes live in separate folders. Your assistant never sees the connections between them.
Memory resets every chat
You explained the architecture yesterday. Today the model has no idea what you're talking about — again.
Your AI, now with a persistent memory layer
Memorytonics turns any folder into a structured knowledge graph your assistant can navigate instead of re-reading raw files every session.
Deterministic AST Extraction
Tree-sitter parses 20+ languages locally — classes, functions, imports and call graphs without touching an LLM
71.5x
Fewer tokens per query vs raw files
~/repo · building graph
Pass 1 · ASTParsing AST with tree-sitter…
Extracting concepts from docs
Clustering via Leiden detection
Emitting graph.json + report
Everything your AI needs to stop re-reading files
Structured extraction, persistent storage, visual exploration and MCP-native delivery — the full memory stack in one install.
✨
Generating summary
• Determining key points...
• Converting to Notion Doc
AST + AI Extraction
Tree-sitter parses code locally, subagents mine concepts from docs — two passes, one unified graph
A
Arthur Sjorgen
00:03
We need to make sure this video hits a little more on a slight angle and...
Interactive Visualization
Explore your graph as a clickable, filterable map — communities, god-nodes and cross-links at a glance
⌘
VideoDocPeople
▶Chizu Brand Design
▶Design Proposal Tem...
People
Evelyn Rose
Arthur Sjorgen
Typed Graph Queries
Traverse connections, shortest-path, filter by confidence — EXTRACTED · INFERRED · AMBIGUOUS — with token budgets
🇪🇸
🇫🇷
🇩🇪
🇯🇵
🇰🇷
🇮🇹
Any Format, Any Language
20+ programming languages, PDFs with citation mining, images via vision, Markdown, DOCX — one drop-in folder
From any folder to a queryable graph in 3 steps
One install, one command, one persistent memory layer. Point Memorytonics at a directory and your AI finally remembers.
0:00:00
Point at any folder
Run mt install inside your AI assistant, then hand Memorytonics a codebase, a research dump, a notes folder — or all three mixed together.
mt build ~/research · extracting graph
▶
Parsing AST · 20 languages…
Extracting concepts from PDFs
Clustering communities (Leiden)
Emitting graph.json + report
The graph builds itself
Pass 1 extracts code structure deterministically on your machine. Pass 2 runs parallel AI agents over docs and images. Every edge is tagged EXTRACTED, INFERRED or AMBIGUOUS so you know what to trust.
Connected assistants
Claude Code
Anthropic
✓
Cursor · Codex
IDE
✓
Gemini CLI · Copilot
CLI
✓
C
G
C
Query from any AI
Share your documented knowledge instantly with your team through simple links. Publish to Notion, Slack, or any platform where your team works together.
Works with every AI assistant you already use
One graph, one MCP server — queried by every model, CLI and IDE in your stack
C
Claude
G
Gemini
✦
ChatGPT
{ }
Codex
▸
Cursor
✦
Copilot
∷
Perplexity
x
Grok
≋
Windsurf
[ ]
OpenCode
T
Trae
A
Antigravity
+ any MCP-compatible client · drop in your own
Use cases
Learn moreNews & intelligence
Track entities, claims and source reliability across thousands of articles.
Legal & case law
Map cases, precedents and statutes into a queryable graph of authority.
Case studies
Build a searchable memory of past projects, outcomes and decisions.
Codebases
Give Copilot, Cursor and Claude the whole architecture, not a 5-file slice.
Product knowledge
Consolidate docs, tickets and Slack threads into a single team graph.
Persistent
AI Memory
01
Index
02
Extract
03
Graph
04
Query
Persistent
Searchable
Token-efficient
Visualized
Multi-modal
Local-first
MCP-ready
Join Thousands of 
Engineers & Researchers
Who Gave Their 
AI 
A Real Memory
·Latest updates
All posts
Build a Knowledge Graph From Your Entire Codebase — Without Sending Your Code to Anyone
knowledge-graph • April 20, 2026
Knowledge Graph for Healthcare Teams — Build Clinical Memory Without Exposing Patient Data
healthcare • April 20, 2026
Knowledge Graph for Lawyers and Law Firms — Build Institutional Memory Without Exposing Client Data
legal-tech • April 20, 2026
Pricing that scales with your graphs
Start local and free. Upgrade when you need hosted MCP, team graphs or enterprise controls.
Save $72/year with annual billing
Solo
For individuals indexing personal repos, notes and research. Runs fully local.
$0.00/month
What's included
1 active knowledge graph
Local AST extraction
Bring your own AI key
MCP server · local only
Interactive graph viewer
Pro
For engineers and researchers running graphs across many projects and assistants.
$72.00/month
What's included
Unlimited graphs
Parallel AI concept extraction
PDF, image & DOCX support
Auto-sync with --watch & git hooks
Hosted MCP endpoints
Cypher / GraphML export
Priority extraction
Enterprise
For teams that need private deployments, SSO and org-wide shared memory.
Custom Pricing
What's included
Everything in Pro
Private VPC deployment
SSO, SAML & audit logs
Shared org knowledge graphs
Custom extraction models
Dedicated solutions engineer
Custom SLA
Give your AI a memory that actually sticks.
One command. Any folder. Every AI assistant you already use — finally on the same knowledge graph.
Knowledge Graph
AI Presets ▾↺↻
▶◁▷🔊00:00:00 / 3:20:10✂⚙⛶
◐□≡≡
Background Image
PresetImageSolidGradient
Background Blur
Image Enhancement
Face Retouch
0:000:150:300:451:001:151:30
Graph · 482 nodes · 1,204 edges