Signal Collective
Hong Kong Book a call
Signal Collective The Library Memory
We find where your growth is · Then we build it

Memory · What your AI knows about you before it talks to you. · a tool

Ask your own notes a question, get an answer

The open-source project khoj, by khoj-ai

“Your AI second brain. Self-hostable. Get answers from the web or your docs. Build custom agents, schedule automations, do deep research.” — the project's own words, on GitHub

What it does
Khoj turns the notes, PDFs, and files already sitting on your computer into something you can actually ask questions of. Point it at a folder of documents and it reads through all of them, then answers in plain language when you ask what you wrote about a client last year, where a number came from, or what you already decided about something. It runs as a private, self-hosted app on your own machine or server, using whichever AI model you already have an account with, so nothing you write leaves your control unless you choose to share it.
Replaces
a paid note-search or personal-wiki subscription
For
A founder with years of notes, PDFs and files scattered across folders who wants to ask them questions instead of hunting through them by hand.
Not for
Not for someone who wants a five-minute mobile app — it needs a machine that keeps the server running, and a first install with Docker or Python before it does anything.
Setup
25 min · Claude Code, Kimi, Gemini CLI or Codex · docker (or python and pip), a machine that can keep the server running, an API key for OpenAI, Anthropic or Gemini, only if you want a hosted model instead of a local one
Notes everywhere One shelf, filed Read before it answers inbox projects decisions it already knows

Get it running

Paste this into the AI that runs on your computer. It does the install, checks it works, and tells you what to type first. If anything fails, paste the error back to it.

Paste into Claude Code
Install khoj (https://github.com/khoj-ai/khoj) on this computer and get it working for me. I am not a developer; explain each step in one line as you go and never skip one. You are Claude Code, running on my machine.

1. Check what this machine already has (git, Python or Node as the project needs, Docker if the README says so). Tell me anything missing and install it, asking me before anything that needs my password.
2. Clone https://github.com/khoj-ai/khoj into ~/tools/khoj and follow the README's install exactly.
3. Configure it. If it needs an API key or a login, stop and ask me for it; never guess one and never store it anywhere except where the README says.
   Tool-specific notes: Use the docker-compose path: mkdir ~/.khoj, cd into it, download https://raw.githubusercontent.com/khoj-ai/khoj/master/docker-compose.yml, then run docker-compose up.
Before that, set KHOJ_ADMIN_PASSWORD and KHOJ_DJANGO_SECRET_KEY to something secure — generate them yourself rather than asking me, unless I want to pick them.
If I want a hosted model instead of a local one, ask me for an OPENAI_API_KEY, ANTHROPIC_API_KEY or GEMINI_API_KEY before setting one; never guess or reuse one.
Setup is done when the logs show "Khoj is ready to engage" — then open http://localhost:42110 and confirm the app loads.
4. Run the smallest test the README gives, and show me the output.
5. When it works, tell me what to type first, in one line, for this job: Feed it last year's notes.
If anything fails, show me the exact error and fix it before going on.

What to point it at first

1

Feed it last year's notes

Point it at a folder of old notes or meeting docs and ask what you decided about a specific client — a good answer quotes the actual line back to you; if it can't find it, that folder isn't indexed yet.

2

Ask a question across scattered files

Ask something that spans several documents, like every mention of one supplier, and check that a good answer names which file each mention came from so you can go check the source yourself.

3

Try one scheduled digest

Set up a single simple scheduled question, like a weekly summary of new notes, and read the first two or three digests in full before you trust it to keep running unattended.

What it must never do unattended

  • Never let it delete a note. It files; you prune.
  • If you configure a hosted model (OpenAI, Anthropic, Gemini) instead of a local one, your documents are sent to that provider to answer — keep anything sensitive on a local model instead.

Who made it

khoj-ai/khoj on GitHub, under the AGPL-3.0 licence. 37,227 stars, checked 9 September 2026. Last change 2 August 2026. We did not write it; we checked it, and wrote this page so you can use it.

Also in Memory

Notes you search Notes your AI reads inboxprojectsdecisionsarchive knowledge write this to my brain
001

Second Brain Kit

Your AI reads your notes before it talks to you, and files what mattered when you're done.

20 min · Claude, ChatGPT or Gemini · a place for notesOpen

What's in the library is how we work.

If you're launching or growing a brand across Asia and the West, a call is where we work out whether it's a fit.

Book a call Back to the library