Plan AI

Appearance

Quick Start ​

The absolute fastest way to start using Plan AI is to use our fully managed, securely hosted cloud platform.

1. The Cloud Platform (Recommended) ​

You do not need to install anything or manage servers to use Plan AI.

  1. Head over to plan-ai.blueberrybytes.com.
  2. Sign in with Google or create an account.
  3. Plug in your OpenRouter and Deepgram API keys in the Workspace Settings.
  4. Download the Desktop Recorder and start capturing your meetings.

It takes less than 2 minutes to get fully set up.

2. Running Locally (For Developers) ​

We’ve designed Plan AI to be incredibly easy to spin up locally. If you want to contribute to the open-source project or self-host the platform on your own machine, follow these steps.

Prerequisites ​

Before starting, ensure you have the following installed on your machine:

  • Node.js (v22.17.1 or higher)
  • Yarn
  • Docker Desktop (For running the local Postgres, Redis, and Qdrant databases)
  • Python (via uv for the Voice AI microservice)

You will also need two API keys:

  1. OpenRouter API Key for LLM processing.
  2. Deepgram API Key for audio transcription.

1. Install Dependencies ​

Plan AI is a monorepo. We use a master script to install all dependencies across the frontend, backend, native apps, and Python microservices simultaneously.

First, ensure you have our code intelligence tool installed globally:

bash
npm install -g gitnexus

Then, from the root of the repository, run:

bash
yarn install:all

2. Environment Setup ​

Generate the necessary .env files for all applications across the monorepo by running the setup script from the root:

bash
yarn setup:env

Once the .env files are created, open plan-ai/backend/.env and insert your OPENROUTER_API_KEY and DEEPGRAM_API_KEY.

(Note: You will also need to configure your Firebase project credentials in these files for authentication to work properly).

3. Start Local Databases ​

Before starting the applications, you must start the local Postgres, Redis, and Qdrant databases via Docker:

bash
yarn docker

Once the database containers are running, push the latest Prisma schema migrations to set up your tables:

bash
yarn db:migrate

(Optional) Seed the Database ​

You can automatically provision an Admin user (admin@plan-ai.local / password123) by running:

bash
yarn db:seed

This grants you immediate access to the Admin Dashboard without needing to manually register via the web UI.

4. Start the Platform ​

From the root of the project, use the following master script to launch everything concurrently:

bash
yarn dev

This single command will simultaneously start:

  1. The Backend API (Node/Express on Port 8080)
  2. The Web Frontend (React/Vite on Port 3000)
  3. The Voice Biometrics Service (Python/FastAPI on Port 8001)
  4. GitNexus MCP Server (For local code intelligence)

Success! You can now visit http://localhost:3000 to log in and start using Plan AI.

Starting the Client Apps (Optional) ​

If you also want to test the native clients alongside the web platform, open a new terminal tab at the root of the project and run:

For the Desktop Recorder:

bash
yarn dev:recorder

For the Mobile App:

bash
yarn dev:mobile