Installation Guide¶

This guide covers the complete installation of PADAI: the backend server (for IT administrators) and the PadaiPanel client (for each RPA user).

System Requirements¶

Client (PadaiPanel)¶

Component	Minimum Version	Notes
Windows	10 (build 1903) or 11	64-bit required
Power Automate Desktop	Current version (2024+)	Must be installed and functional
.NET Desktop Runtime	10.0	Required for PadaiPanel
Disk space	500 MB	For the panel and local data
RAM	4 GB	Minimum recommended

Server (IT Administrator)¶

Component	Minimum Version	Notes
Python	3.11+	For the FastAPI backend
Docker + Docker Compose	20.10+ / 2.0+	Alternative to direct Python deployment
OpenAI or Azure OpenAI API key	--	Provided by the client (BYO Key)

Bring Your Own Key

PADAI uses a BYO Key model. You provide your own OpenAI or Azure OpenAI API key. AI API call costs are billed directly by your AI provider, separate from the PADAI license.

Server Installation (IT Administrator)¶

Choose one of two deployment options: Docker (recommended) or Python direct.

Option A: Docker Deployment (Recommended)¶

1. Prepare the Bundle¶

The client bundle contains:

padai-api.tar -- Docker image
docker-compose.yml -- service configuration
license.json -- signed license file
.env.example -- configuration template

2. Load the Docker Image¶

docker load -i padai-api.tar

Note

No internet download is required. The deployment works entirely offline once the image is loaded.

3. Configure the Environment¶

Create a .env file next to docker-compose.yml:

# API Key (choose OpenAI OR Azure)
OPENAI_API_KEY=sk-...

# For Azure OpenAI, comment out the line above and use:
# AZURE_OPENAI_API_KEY=...
# AZURE_OPENAI_ENDPOINT=https://<instance>.openai.azure.com
# AZURE_OPENAI_API_VERSION=2024-02-01

# Authentication
API_TOKEN=your-secret-token

# License
LICENSE_SECRET=your-hmac-secret
LICENSE_REQUIRED=true
LICENSE_EXPIRED_MODE=block

# Rate Limiting
RATE_LIMIT_ENABLED=true
RATE_LIMIT_PER_MIN=120

# Persistence (true = save generated plans)
PERSIST_CONTENT=true

4. Place the License File¶

Copy license.json into the same directory as docker-compose.yml. It is mounted as read-only automatically.

5. Start the Server¶

docker compose up -d

6. Verify¶

# Health check
curl http://localhost:8000/api/healthz
# Expected: {"status":"ok","license":"ok","version":"..."}

# Test generation (replace $API_TOKEN with your token)
curl -X POST http://localhost:8000/api/plan \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $API_TOKEN" \
  -d '{"prompt": "Display a Hello World message"}'

Option B: Python Direct Deployment¶

1. Extract the Project¶

Extract the provided archive to a directory of your choice.

2. Create the Python Environment¶

cd ia-pad-agent
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

3. Configure¶

Create a .env file in ia-pad-agent/ with the same variables as Option A (see above). Place license.json in the same directory.

4. Start the Server¶

python -m pad_server.main

The server starts on http://localhost:8000.

5. Verify¶

curl http://localhost:8000/api/healthz

Client Installation (RPA User)¶

1. Obtain PadaiPanel¶

Download PadaiPanel.exe from the distribution channel provided by your IT team, or install from the Microsoft Store (User plan).

2. Install .NET Desktop Runtime¶

If .NET 10 Desktop Runtime is not already installed, download and install it from the official Microsoft website. PadaiPanel will warn you at launch if it is missing.

3. Configure the Server Connection¶

At first launch, PadaiPanel displays an onboarding wizard. Enter:

Field	Value	Example
Server URL	Address of the PADAI backend	`http://padai-server:8000`
API Token	Authentication token provided by IT	`your-secret-token`

These settings are saved locally in %LOCALAPPDATA%\PadaiPanel\.

4. Launch PadaiPanel¶

Open Power Automate Desktop and open a flow in the Designer.
Launch PadaiPanel.exe.
The panel automatically docks to the right side of the PAD Designer window.
The connection indicator at the bottom of the panel confirms the server connection.

Tip

PadaiPanel automatically detects open PAD Designer windows. If multiple windows are open, a separate panel instance is created for each.

Network Configuration¶

Ports¶

Port	Direction	Usage
8000 (TCP)	Client to Server	PADAI REST API
443 (TCP)	Server to Internet	OpenAI / Azure OpenAI API calls

Firewall Rules¶

Server: open port 8000 inbound for client workstations
Server: allow outbound connections to api.openai.com (OpenAI) or <instance>.openai.azure.com (Azure)
Client: allow outbound connections to the PADAI server (port 8000)

Proxy¶

If an HTTP proxy is configured on the network:

Server: set HTTP_PROXY / HTTPS_PROXY environment variables
Client: PadaiPanel uses Windows system proxy settings automatically

HTTPS (Recommended for Production)¶

For production deployments, place a reverse proxy (nginx, Traefik, IIS) in front of the PADAI server with a TLS certificate. Set ALLOWED_ORIGINS in the server .env with the HTTPS URL.

License Management¶

How Licensing Works¶

PADAI uses a license system based on signed JSON files. The license controls:

Maximum number of users (max_users)
Maximum runs per month (max_runs_per_month)
Enabled features (features)
Expiration date

Installing a License¶

Receive the license.json file from PADAI Technologies.
Place the file in the server directory.
Configure environment variables: LICENSE_FILE, LICENSE_SECRET, LICENSE_REQUIRED=true.
Restart the server.
Verify: GET /api/license/status returns the status, quotas, and expiration date.

Expiration¶

When a license expires with LICENSE_EXPIRED_MODE=block, code generation is refused (HTTP 402). Read operations (viewing runs, checking status) remain available. Contact PADAI Technologies for renewal.

Next Steps¶

Once installation is complete, proceed to the Getting Started guide for your first AI-assisted code generation.