Towels, a Roomba, dinner, and translation

I saved about $400 by dyeing my towels instead of replacing them

My towels were faded but still good. I photographed them, showed AI the colors I wanted, weighed each batch, used test cloths, and corrected the calculations as the plan changed.

What I got out of it: The finished colors worked, we kept the towels, and I avoided spending about $400 replacing them.

Why AI was useful: It could work from my towels, measurements, test results, and color choices instead of giving me one generic dye recipe.

I researched replacement wheels for my Roomba

The rear wheels on our nearly five-year-old Roomba j7+ were almost worn smooth. I supplied photos and used AI to identify likely OEM and aftermarket replacement parts within my price limit.

What I got out of it: A reasonable repair path that may let me keep the vacuum. I have not installed the parts yet.

Why AI was useful: I could start with pictures of the actual wear. I still had to correct it when it confused the j7+ with a Combo j7+.

I made naan pizza from what was actually in the house

We had naan and garden vegetables. The first AI suggestions called for ingredients I did not have, so I redirected it toward balsamic, ranch, honey mustard, and the rest of the actual refrigerator.

What I got out of it: A dinner plan built around what was available instead of another grocery trip.

Why AI was useful: I could keep correcting the suggestions until they fit our kitchen rather than search for a recipe that happened to match it.

I turned nine produce photos into a usable list

I photographed the garden produce and asked AI for a recipe-friendly inventory. It called roughly fifteen lemon cucumbers potatoes. I corrected it, and it rebuilt the list.

What I got out of it: An organized starting point without identifying everything one item at a time.

Why AI was useful: I could give it the whole pile at once, keep the confidence limits visible, and correct the miss.

I translated a DFO document and checked it on the way back

For a Hands On AI Lab class, I translated a Dear Future Overlords document into conversational Mexican Spanish, then blind-translated that version back into English to check whether the meaning held together.

What I got out of it: A practical translation exercise I could inspect from the language I know. A class participant was impressed by the result.

Why AI was useful: I could create the translation and run a meaning check without pretending I could personally judge every Spanish phrase.

None of this was especially futuristic.

I saved some money.

I avoided a couple of extra errands.

I used things I already had.

I fixed a few small problems faster.

That is mostly what AI looks like in my life.