Shaarav4795

I fixed the bug where changing a setting required the app to relaunch. I then rearchived and shipped the DMG.

I made the README for this which includes problem statement, target audience, and who its for. I also squashed some bugs and generated the direct download dmg which I will publish as a gh release!

While further testing the app, I noticed that CPU usage would rise to 40 percent, which is simply unacceptable (at least imo). So I spent this time brainstorming and implementing solutions until I finally got the usage to hover around 10 percent, which is way better. In classic mode (where it doesn’t listen to your voice), it’s even lower!!! This means you get more CPU for your other apps :)

I finished the website for PromptMe! I tried to make it look really nice, however I didn’t spend THAT much time on it since I was getting bored, lol… I attached a few screenshots, added a nice hero message, and here we are… Honestly, I really like this!!!

I designed and made a settings window for the app!
Before, the settings window that I did make was really basic, with just an option for changing between the two modes classic and voice hold.
Now, the user can create and use multiple prompts (scripts), and tweak exact behaviour.
Overall, I feel this is a very good UX change.
Additionally, I dropped in the app icon.
Side Note: I kinda forgot to make a devlog around the 3.5 hr mark, which is when I finished the settings menu and app icon, so he extra one hour includes me starting the website! :)

I added a voice mode where speaking scrolls and silence pauses. The whole point is so you don’t have to click or hover, and that you can speak at your own pace :)

It uses the built in macOS speech recognition as well as audio signal detection. Speaking makes it scroll, silence makes it hold. I also built a live signal bar chart so you can actually see it in the overlay.

The transcriptions get matched word by word against your script so it knows roughly where you are, even if you paraphrase or skip ahead, and it highlights your current line to guide you.

It took a lot of time to make it less jittery and more stable, and in noisy places it sometimes fails, but overall its quite nice and I hope to improve it a bit more in my final polish…

Today I wrapped up Phase 1 of PromptMe, my notch teleprompter for macOS. The app is now fully functional as a menu bar utility that sits hidden from the Dock and overlays your script right under the notch.

The overlay window has a custom notch shaped design with controls at the top for play, pause, jump back, speed adjustment, and opening settings. The auto scroll engine has hover to pause, a countdown before start, and a 5-second rewind button.

Problems and Bugs: CPU usage during active scrolling is a lot, around 40% (not where I want it). My next code will add voice assisted prompting so the scroll follows your speech instead of a fixed speed, as well as (trying!) to bring the CPU usage down to say, 15%.

PS: Sorry, in the video you cannot see my cursor for some reason, so it looks like it randomly pauses, but that is the hover to pause feature!

I cleaned up the code way more… I know I’ve said this before, but previously I just moved a few components here and there and called it a day. However, I did a full refactor and meticulously moved every single component, well, not every single one, obviously - I don’t have time for that. I also cleaned up some obscure references to my old project name, so I just wanted to migrate everything to the new ezMail name that my project uses. Nothing much else, it’s just a bit of polishing here and there to get it ready, because I’m going to ship next.

I added a demo video to the site and added optimisation for server side rendering for CLoudflare Pages which is what I will use to host the site.

I cleaned up some of my code and put lots of things into their own components. I also finished writing the README for this project, which includes the problem statement and target audience (my dad totally did not tell me to put this in).

I enhanced the templates quality and made them more distinct (because before they were basically the same except for different text). Overall, this makes it feel so much more polished, and the website truly feels complete now :)
Now I just need to make this ship ready and polish the polishings…

I added the export page. Here, the user can see a final live preview and export it for sending, by either copying, downloading it as .eml (a type of file format for emails), or pressing the open in mail button. By far the most difficult thing was getting the copy button to work as I had to research only to find out that you don’t copy HTMl, you copy DOM (which is basically the rendered part). However, I’m super happy with this screen, and my next job will be to improve the quality of the templates.

I completely implemented the editor page. The editor page opens when you click start building and allows the user to edit the text in the template, and change fonts, colours, and text effects using a floating top bar, similar to Canva. It took a lot of time to get right because sometimes images wouldn’t load, and on smaller devices it was very cramped. But finally, after 6hrs (😅) I’m done… at least with this page :)
(AI used in this: the text for the emails and actually designing HTMl email templates as there are so many restrictions)

I made the template gallery look MUCH better by adding previews, adding an adaptive grid layout, and filters on the side. I also coded an algorithm that dynamically adjusts the grids for a better experience.

I designed the landing page for the website. I spent a lot of time on this because I wanted it to look STUNNING!! I used a greyscale colour scheme as that is my personal preference. I also scaffolded a template library that will use a grid design. It’s not 100% done yet but I should be able to finish it in an hour or so…

I made this ship ready by making a README and modifying a website I had already made to suit the app (maybe because I didn’t feel like making yet another website lol)
I also added the Github Release in the form of a compressed DMG (as for some reason normal DMG upload wasn’t working…)

I added an onboarding flow for easier setup. I also optimised the app… again. This was the worst part, as I thought I was done but no. The app loves hogging CPU. Although the fix was easy, just not making it monitor as much in the background and dynamically scaling FPS and the app is back to sub‑1 percent at idle.

I improved the settings menu a lot. Before, it was just a simple runner selector, but now I’ve stylized it, made it easier to understand, and added many more options. The user can change the dashboard tiles, change the runner (plus a search bar), launch at login, randomly change the runner, and much more. I’m so happy with this, as it makes the UI and UX much better. I also added the app icon (the layers with Canva, put together in Icon Composer).

I designed and made the popup for when you click on the menu bar icon. it shows your CPU load, memory usage, breakdown of your CPU cores, uptime, a graph of your usage history, and a a lsit of the top 5 processes. It also has a quit button and a settings button. Overall I’m really happy with how this looks.

I optimised the app WAYYY more. This part was more difficult than before because the bulk of optimisation was done; it was just the final things that would bring it down to ~1%. Most of this time was spent staring at activity monitor until I found something wrong and then fix the bug. But now, the result is less than 0.9% CPU usage and less than 50MB memory, which are both equal or better than RunCat!!!

I wrote the code for the app to create the MVP. To make this, I had to:

1. First extract the assets of the RunCat avatars using the .car library.
2. Write Swift code that selects the Puppy avatar (because that’s my favorite) and animates it faster the higher the CPU usage.
3. I looked at Activity Monitor and, oh god… my app was using 20 percent CPU. 20 percent!!!!! I urgently quit the app and immediately capped the FPS and applied some basic optimisations, which brought it down to about 7 percent.
4. I still wasn’t happy with that, so I asked ChatGPT for help and then made the following improvement: the app keeps each runner animation in a small cache, which avoids repeated frame reloads, and it only updates the menu icon when the image actually changes.

All together this brings CPU usage down to around 3 percent, which is good but I’m not 100 percent happy with it so I will keep optimising!!

Implement cloudkit to replace supabase, as well as cloudkit syncing across devices

I added a Supabase intergration so users can share studysets. there is also link sharing but that requires and apple developer account.

Update website to look wayyyy better. Have a few placeholder images, I’ll replace them once I get the app on an actual device.

Added better notifications to make it more personal and motivational.

Implemented way better Spaced Repetition Scheduling into the app to have a better UX

Finally fixed the lag problem for REAL! This took way too much time…

Allow user to manually edit flashcards/quizzes and fixed a few lag problems.

Add better rate limit handling, better fallback models, update warning message in settings to communicate this better.

Add micro animations
Add spaced repetition algorithm
Add better dark mode styling
Squash some bugs

Just built v1 of the app - now fixing on fixing bugs and making the UI better!