id: FoafZtNN
recording_id: FoafZtNN
audio_file: audio/FoafZtNN.mp3
parent_id: null
title: Save My Whiteboard Android app for Drive-backed photo sessions and AI outputs
recorded_at: '2026-04-26T11:01:43.000000Z'
created_at: '2026-04-26T11:01:48.000000Z'
updated_at: '2026-04-26T11:02:11.000000Z'
duration: 214343
public_slug: null
is_published: 0
recording_type: 1
user_id: 131594
user_name: Daniel Rosehill
tags:
- Tech
- AI
- Idea
- Android
- Data
- Work
Save My Whiteboard Android app for Drive-backed photo sessions and AI outputs
I want to create an app. It'll be called Save My Whiteboard. Very simple Android app that will work as follows: user saves a whiteboard or multiple photos; it's like a session. So in other words, when a whiteboard is photographed, there might be slightly different angles of the same whiteboard, but the assumption is that it's the same one in the same whiteboard.
What I've used before is using Gemini 3.1, a workflow for creating a tech diagram from a whiteboard that's kind of my messy handwriting, saving of course the original whiteboard, running a pass that fixes white balance issues. Sometimes if I take it to my office, it comes out yellow-tinted, and a simple white balance correction will make it much more pleasant to look at.
Another one would be cropping the whiteboard automatically, even cropping out the physical frame of the whiteboard. Then those are image-to-image tasks, and image-to-text would be transcribing the whiteboard, denoting the whiteboard, and those I haven't really explored yet. But the idea I have for this app or utility is user defines a folder, a base folder, in their Google Drive, and then it creates folders with date-based structures.
So if today is the 26th of April, the folder is 2604. The user takes photos, hits Save, and number one, we create a subfolder that has the timestamp. An alternative organizational approach is just to have one folder which records (I think this might actually be better) date_ddmmyy_mmhh_timestamp, in other words.
Then the user selects what they want to do. Do they want a text pull-out? Do they want a tech diagram? Basically, what would happen is that the user would hit Upload. This gets sent to Gemini, and the transformations the user has requested will be applied. So we'll always firstly save the originals in a subfolder called Originals.
Then we'll have a subfolder called Generated Images, which is the image-to-image. So it might be the tech diagram, then one called Text Extracts. The idea is to have a simple camera app on your phone that will allow you to take a photograph of a whiteboard, send it to the cloud.
I think the only split-out that I'd like is always save the originals. So even if all the generative AI stuff fails because we're going to need to put an API key for Open Router, at least we always have those originals. So the user can then wipe the whiteboard, and they know that it's being preserved, and they know where just as importantly, they know where that data has gone in the cloud.