diff --git a/README.md b/README.md
index 9e88713..bce5ab8 100644
--- a/README.md
+++ b/README.md
@@ -39,7 +39,7 @@ mini_page/
 1. Flask renders `index.html` with the appropriate translation set, selected via the `?lang=` query parameter (`it` by default, `en` also supported).
 2. The page displays a banner and four clickable cards. Each card click triggers a JavaScript modal overlay that loads one of three content types:
    - **Video** — plays a local `.mp4` file via `<video>`
-   - **Iframe** — embeds the GP widget served at `http://localhost/viewer`
+   - **Iframe** — embeds the GP widget
    - **PDF** — displays a local PDF via `<iframe>`
 3. All static assets (images, videos, CSS, PDF) are served directly by Flask from `src/resources/`.
 
@@ -49,10 +49,9 @@ mini_page/
 
 - **Python >= 3.13**
 - **[uv](https://docs.astral.sh/uv/)** package manager (strongly recommended)
-- **Docker** — required to run the GP widget locally (stable release **v0.1**)
-- **Internal VPN** — must be active for the GP widget (Phenological Recognition card) to connect to backend services
+- **Internal VPN** — must be active for the GP widget (Phenological Recognition card) service
 
-> The GP widget is expected at `http://localhost/viewer`. Without Docker and an active VPN connection, Card 1 (Phenological Recognition) will not function.
+> Without an active VPN connection, Card 1 (Phenological Recognition) will not function.
 
 ---
 
@@ -64,24 +63,13 @@ mini_page/
 uv sync
 ```
 
-### 2. Start the GP widget via Docker
-
-> **VPN must be connected before this step.**
-
-```bash
-docker run <gp-widget-image>:v0.1
-# Replace <gp-widget-image> with the actual image name for GP widget v0.1
-```
-
-The widget must be running and reachable at `http://localhost/viewer` before opening the page.
-
-### 3. Start the Flask server
+### 2. Start the Flask server
 
 ```powershell
 uv run python src/main.py
 ```
 
-### 4. Open the application
+### 3. Open the application
 
 | URL | Language |
 |-----|----------|