Had one last day of moving

Made progress on algo #tutoic

made progress on algorithm #tutoic

continued work on algorithm #tutoic

Added some tables that are the bases of the high level system of the new algorithm #tutoic

implemented frontend side of lower system algo #tutoic

Built new algo in frontend, start testing / debugging #tutoic

Started some work on higher level system of learning algorithm #tutoic. The planned algorithm is the right one because it can be very efficient with memoizing, but now I'm just implementing the simple lazy version so I can work it into the app and validate. Didn't build much but tomorrow there's no day job so will build much more.

Made a really nice breakthrough in the learning algorithm. The design is now finished in my head, all that's left is to execute. It's made of two systems: the high-level system and the low-level system. I've finished implementing the low-level system today. What's left is high-level and integration into the app. Implementation-wise the last of those three will be the most time-consuming. #tutoic

solved that i downed one of my apis on a release yesterday & added a dedicated fact page to more easily browse around knowledge data & better interpret the algo's choices while i'm developing it

Made progress on teaching algorithm #tutoic. This is a tough nut.

Fixed a few minor issues in login/logout mechanics and some of the editors of #promptgrid. Dayjobday & renovations happening at home, so likely won't be able to do more today.

Improved the example content new #promptgrid accounts get upon registering

#promptgrid now populates any new user's account with an example prompt with example test inputs. Will add example evaluators later but have to stop now.

Today: finished those #promptgrid breadcrumbs that I started this morning; finished an update of #tutoic's to Vite and the most recent React which I started earlier this week (honestly, shouldn't have started it... major waste of time at this moment); fixed a bug that caused analytics events for #promptgrid to not arrive in GA; added a simple explanation of how to consume one of the #promptgrid prompts in your own applications. #promptgrid is still an impossibly shit experience for new users but... getting there. Good night fellow builders 🌙

Woke up 4:30. Fixed that analytics events weren't coming through well in Google Analytics for #promptgrid . Started adding breadcrumbs; doing a global breadcrumb bar like Vercel has, I like it because it's spatially efficient.

Made a landing page with a very simple "AI prompt generator", with a single text box to accept user input. This landing page is to capatalize on the search terms where I was getting the best Google Ad traffic, while gathering user input to learn about their problems. It's a bit of a pivot from #promptgrid core idea but I'm experimenting with being more flexible. And maybe I can upsell users to the core idea.

Just finished building the test runner for my new product, a prompt engineering web app. It is now officially usable by myself. Todo: start using it for my own apps & make UI self-explanatory so users besides me can use it. The first Google Ad (budget €5 per day) meanwhile still on zero impressions, third day after start. Ad is rated "good" in Google Ads dashboard. Landing page shit?