Quick start

Six steps from free trial to an AI that remembers.

On this page

  1. Get your free trial license
  2. Check your email
  3. Install and activate
  4. Set up and start the daemon
  5. Connect your AI — orkai mcp-config
  6. You're ready

1. Get your 7-day free license

orkai requires a license even for the trial. Start here — no subscription, one device, $0 checkout.

  1. Open Pricing and click Start 7-day free trial.
  2. Complete checkout on Polar (merchant of record).
  3. Your license key arrives by email shortly after checkout — continue with step 2 once it lands.

Prefer to buy outright? Choose Single device ($49.99) or Three devices ($99) on the same page — the steps below are identical, only the key type differs.

Start 7-day free trial

2. Check your email — license key and download

Polar sends a receipt email within a few minutes. That email contains your ORKAI_ license key and download instructions for your platform.

Note: Do not install anything until this email arrives — the download link and activation key are sent together after checkout.

Didn't receive it? Check spam, then email support@getorkai.com.

3. Install orkai and activate on this machine

Use the download instructions from your Polar email. Then activate — device cap depends on your plan (1 or 3 machines).

Download the binary for your platform from your Polar receipt email. Official package managers (Homebrew, winget) ship in a later distribution milestone.

macOS

Download the orkai binary from your Polar receipt email (or GitHub Releases when published).

Make it executable and move it onto your PATH:

chmod +x orkai
sudo mv orkai /usr/local/bin/

Linux

Download the orkai binary for your architecture (amd64 or arm64).

Make it executable and move it onto your PATH:

chmod +x orkai
sudo mv orkai /usr/local/bin/

Windows

Download orkai.exe from your Polar receipt email (or GitHub Releases when published).

Place it in a folder on your PATH (for example C:\orkai\) and add that folder to your user PATH environment variable.

orkai version

Activate your license

After install, activate with the key from your email:

orkai activate ORKAI_YOUR_KEY_HERE
  • Your license key arrives by email from Polar after checkout. Keys normally start with ORKAI_.
  • Device cap depends on plan: trial and single-device = 1 machine; three-device = 3.
  • Upgrade from trial: run orkai activate with your paid key; it replaces the trial in ~/.orkai/license.json.

Verify with orkai license status — it should show your plan tier, device cap, and expiry/grace state.

4. Set up and start the daemon

The first time you run orkai, an interactive setup walks you through local configuration — embedding provider (OpenAI, Ollama, or bring-your-own vectors), storage paths, and credentials. This writes ~/.orkai/config.yaml and your local data directory.

First run only — interactive setup

orkai serve
# runs in the foreground — logs stream in this terminal
# completes the first-run wizard, then keeps running until you Ctrl-C

Requires an interactive terminal. Without activation from step 3, orkai refuses to start.

Every day after — keep the daemon running

orkai start    # background daemon — stays up on macOS/Linux
orkai stop     # stop when you need to
orkai status   # confirm it is running

Prefer start over serve for normal use. serve ties up a terminal with live logs; start keeps orkai running in the background while you work. Use orkai logs -f when you want to tail output without blocking a shell.

  • Live ports are written to ~/.orkai/runtime.json when auto-fallback occurs.
  • Open the local UI with orkai open while the daemon is running.

5. Connect your AI — orkai mcp-config

With the daemon running (orkai start — or orkai serve while debugging), print a ready-to-paste MCP block: 

orkai mcp-config

Copy the JSON into your assistant's MCP settings. The block includes the SSE URL and bearer token for your local daemon.

Cursor

Open Settings → MCP. Add the orkai server entry from orkai mcp-config. Restart Cursor if tools do not appear immediately.

Claude Desktop

Paste into claude_desktop_config.json under your user config directory. Merge the mcpServers.orkai block — do not replace other servers.

Claude Code, Windsurf, and other MCP clients

Same JSON shape: SSE URL plus Authorization: Bearer …. Paste wherever that client expects MCP server definitions.

ChatGPT and other assistants

If your assistant supports MCP connectors, use the same output. Look for an MCP or custom connector section in that app's settings.

Note: Complete first-run setup with orkai serve once, then use orkai start before running orkai mcp-config.

6. You're ready

Open your AI assistant and start a session. A well-configured agent calls overview() first — recent sessions, standards, skills, and tools scoped to your project.

Your agent can now save documents, preferences, and session summaries on your machine — memory that persists across chats.

Optional — index a folder

Point orkai at any local directory — a code repo, research folder, analyst workspace with CSVs and PDFs, or notes on your Desktop. See Projects & indexing. 

cd /path/to/your/folder
orkai init
orkai index .              # code + documents + analytics (default)
orkai index document .     # documents only
orkai index analytics .    # spreadsheets only

Next reads