I built a food tracker with AI because I didn’t want to pay for one that wasn’t quite right for me. It took maybe a couple of hours of my own time to get it into a state I was happy to use.
I agree with Pieter Levels that AI will take over tasks people currently use separate apps for. I think it will replace most SaaS products, starting with simple personal apps and moving toward larger business systems.
With my tracker, I just tell AI what I have eaten and what should be logged. One lunch was leftovers, about 200 grams of mashed potatoes, 150 grams of minced meat, some mozzarella, and dark chocolate afterward. It can look up published nutrition values and estimate portions when I haven’t weighed the food.
Getting the dashboard right was probably the hardest part. My first description wasn’t detailed enough. I had to clarify the weekly and monthly views, and what belonged on the main graph versus a detailed view. I could ask for changes until the screens fit how I wanted to use them.
For businesses, I expect AI to help build an alternative to the SaaS product they would otherwise buy. They still need software, but they can have it built around their own processes. The software doesn’t need AI for every operation.
Shopify’s B2B checkout, for example, doesn’t support local delivery, although B2B draft orders created in the admin do. A business offering wholesale delivery has to work within that distinction. Shopify documents it here.
I expect AI-assisted development to reduce compromises like that. As AI improves, I expect more people to be able to describe what they need and have it built, including people who aren’t technical.
Maintenance is part of that too. I expect AI to follow how a system is used and propose changes or bug fixes. A person approves each change before it goes live.
The SaaS products I expect to succeed will need to solve really hard problems. Something simple enough for a customer to recreate with AI will be harder to sell. I think there will still be room for focused products, possibly in specific niches, where the problem itself is difficult to solve.