The workflow integrates several tools:

@n8n.io
@notion.com API
@github.com API
@langchain.bsky.social
Google Gemini
GPT4o

Till next time, keep building

The resulting documentation is appended directly to the same Notion page the button was clicked on, about 30seconds later.

An AI-driven content generation step, using models like Google Gemini for huge file diffs and GPT-4o, transforms the processed data into clear, instructive documentation. It follows a predefined template for consistency.

Sidenote: The Output-fixing parser from LangChain is 🤯

Once the Notion page details arrive in n8n, the workflow retrieves key pull request details such as the repository name and PR number.

It then combines the GitHub PR DIFF files in a JSON object, filtering out unnecessary files (e.g., package-lock.json) to focus on essential content.

The process begins when a user clicks a button in Notion. This button is linked to both the relevant @notion.com tion page and the corresponding pull request, and triggers an @n8n.io workflow

Update: 3D print prototype complete

Further automation opportunities:
- Add a relay to the D1 Mini to switch on and flash a green or red light
- Create a sensor reading in Home Assistant to trigger/control other items integrated in the house
- Close the garage door if the car is parked (risky)
- Start some music in the living room

Problem solved ✅

UAT Testing. 65cm is my number

TODO: 3D print a case

🖥️ Programming the D1 Mini
1. Connect the D1 Mini to your computer via USB.
2. Go to Home Assistant > ESPHome > New Device.
3. Name it: parking-sensor.
4. Select your USB serial port (e.g., USB Serial (cu.usbserial-210)).
5. Load your config and install

🔌 Ultrasonic Sensor Wiring - This sensor measures distance using sound waves:
Vcc to 5V on D1 Mini
Trig to D3/GPIO0
Echo to D2/GPIO4w
GND to GND

📟 Display Wiring - This display will show the distance readings from the sensor:
Clk to D6/GPIO12 on D1
Dio to D7/GPIO13
Vcc to 5V
GND to GND

Tools:
- Home Assistant
- ESPHome

Parts:
- Ultrasonic Sensor
- 4 Digit LED Display
- D1 Mini

Yes. If you’re running with Docker, you can create your own Dockerfile with all the packages you want, then build and run

Here’s my docs for the ffmpeg package I used

I’ve got my team on Business Standard ($12/mo) that has built in transcriptions and recordings

Main reason is most of them (last time i looked) don’t support custom webhooks so i can own my prompt and output

Thanks to @thomasjfrank.com for inspiring this a couple of months ago with thomasjfrank.com/how-to-trans...

Recent additions: Automatically upload and summarise every business call, and ask in Telegram whether to send it to Team notes or Private Notes.

Now after each call I get the option to send it to the relevant Notion database

The Game Changer: Ask Notion AI about my or my team's action items for the day

Transform and Notify: It converts this data into a Notion page, complete with rich text, to-do lists, and user mentions for immediate task tracking and notifications.

The users in the workspace are prefetched for mentions.