Local knowledge base
Use Flint as a private digital brain on your computer.
Flint is a local-first desktop knowledge base. It gives you Markdown notes, wiki-style links, tags, daily notes, a visual graph, an infinite canvas, and an optional AI agent that can use your notes as context.
The clean setup is simple: install Node.js, Python, Git, and Ollama, run the Flint installer, pull one Ollama model, then verify that the Flint AI agent can see the model.
What you are setting up
Requirements
- Node.js 18 or newer.
- npm, which installs with Node.js.
- Git.
- Python 3 if you want the AI agent.
- Ollama if you want local model chat.
Flint can still be useful as a local notes app without AI, but Ollama is the recommended path if you want private local answers from your own vault.
Windows install
- Install Node.js 18 or newer from the official Node.js website.
- Install Git for Windows and keep the option that lets Git work from PowerShell.
- Install Python 3 if you want to use the Flint AI agent.
- Open PowerShell and confirm the tools work.
node --version
npm --version
git --version
python --version
- Run the Flint PowerShell installer.
irm https://raw.githubusercontent.com/Chintanpatel24/flint/main/install.ps1 | iex
After installation, open Flint from the Start Menu or run flint from a new terminal if the installer added it to your user PATH.
macOS install
- Install Homebrew if you do not already have it.
- Install Node.js, Git, and Python.
brew install node git python
- Confirm the versions.
node --version
npm --version
git --version
python3 --version
- Run the Flint installer.
curl -fsSL https://raw.githubusercontent.com/Chintanpatel24/flint/main/install.sh | bash
After installation, open Flint from your app launcher if it appears there, or run flint from Terminal.
Set up Ollama
- Install Ollama for your operating system.
- Open a new terminal and confirm Ollama works.
ollama --version
- Download a local model. The Flint README gives
llama3.2as the example model.
ollama pull llama3.2
- Start Ollama if it is not already running.
ollama serve
Ollama normally listens at http://localhost:11434. Flint's local agent checks that address and lists the models available on your machine.
Verify Flint AI connection
- Start Ollama and make sure
llama3.2is downloaded. - Open Flint.
- If you are using the agent manually, start it from the terminal.
flint-agent
- Open the Flint AI panel and ask a small test question.
Summarize my current note in three bullet points.
If Flint says the model is disconnected, check that Ollama is running, then run ollama list to confirm the model exists locally.
First productivity workflow
- Create a note called
Inboxfor quick thoughts, links, and tasks. - Create a note for one active project, such as
Website Redesign. - Link notes with
[[Project Name]]so Flint can build a graph. - Add tags like
#todo,#idea, or#meeting. - Use daily notes for work logs and quick summaries.
- Ask the AI agent to summarize, find related notes, or clean up a rough note.
This gives you a repeatable system: capture quickly, connect related information, review the graph, then use AI to turn scattered notes into next actions.
Useful Flint habits
Final checklist
- Node.js is version 18 or newer.
- npm works from the terminal.
- Git works from the terminal.
- Python 3 is installed if you want the AI agent.
- Flint opens as a desktop app.
- Ollama is installed and running.
ollama listshows your model.- The Flint AI panel can answer using your notes.