I built a local-first "Second Brain" with AI chat, knowledge graphs, and Obsidian-like vaults — all running locally on your machine

Ahmed Didat — Tue, 30 Jun 2026 08:28:44 +0000

I've been building something I've wanted for years: a truly local-first "Second Brain" that combines the best of Obsidian, AI chat, and knowledge graphs — all running on your own machine.

I've tried everything: Obsidian, Notion, Roam, Logseq, Tana, Capacities, Mem.ai, Reflect...

The pattern: Either it's cloud-dependent (privacy risk, subscription, offline broken), or it's local but limited (no AI, no knowledge graph, poor mobile).

I wanted both: full local control + AI that actually understands my notes + knowledge graphs that actually help me think. something like "obsidian + AI with RAG" with some other features.

I'm seeing second brain AI frameworks on demand in instagram. People are creating second brains by connecting claude code with obsidian... so i thought what if i bring the whole thing into a software.... And that is the core idea of my project..

✅ What Works Today

Feature Status

Local Vault System ✅ Markdown files in folders you control (Obsidian-compatible)

*AI Chat with RAG *✅ Hybrid search (FTS5 + vector embeddings) + your notes as context

*Knowledge Graph *✅ Auto-extracted entities/relationships, interactive graph

*YouTube Ingestion *✅ Paste URL → transcript → structured markdown notes

Daily Reflections ✅ Auto-generated daily summaries from your activity

*Proactive Suggestions *✅ AI surfaces relevant past notes while you write

*Hybrid Search *✅ keyword + vector database

Local-First ✅ You can use local LLMs, or you can set an API key...

"I wanted a thinking partner, not a filing cabinet."

Most note apps are passive storage. Second Brain is an active thinking partner:

You write → it embeds + indexes
You chat → it retrieves relevant context (hybrid search)
It responds → with citations to your notes
It suggests → "Hey, you wrote about this 3 months ago — relevant?"
It visualizes → Knowledge graph shows connections between ideas and informations that you missed

I need honest feedback before investing more months.

Would you use this? What's missing for your workflow?
Local-first vs Cloud sync — how important is cloud sync vs. pure local?
AI Integration — OpenRouter (cloud) vs Ollama (local)? Both?
Pricing model — Free/OSS core, paid cloud sync? One-time license? Donations?
Plugin system — What extensions would you build?

DEV Community: Ahmed Didat

I built a local-first "Second Brain" with AI chat, knowledge graphs, and Obsidian-like vaults — all running locally on your machine