AI and Personal Software Adapters

There are polarised takes on AI usage. Some very well-known engineers are advocating adoption and others rejecting it all as slop, many of whom I have respect for regardless of what side of the divide they are on.

For me, it’s the difference between a supermarket’s factory-designed bakery churning out loaves of bread for the masses vs the local artisan sourdough baker.

Both are hugely important and have different markets. The regular loaf isn’t what you want slightly toasted with some balsamic vinegar, olives, nice cheeses and a bottle of red in front of a log burner on a weekend evening, right? But for a simple lunchbox cheese and pickle sandwich it’s perfect.

AI is the same. I am finding that I value more indie websites and bespoke handcrafted UIs on the artisan side, the beauty and craft of it, for looking at and reading, but for my own personal use I just need a loaf. I just want a simple slice or two to make the sandwich work.

AI is great for personal software. Not in the sense of the so-called SaaS apocalypse replacing the need for Jira, GitHub and the like with everyone cranking out all sorts of big solutions, but something simpler.

I’m calling them personal adapters. I have been able to use coding agents, where my own mind is failing, to develop solutions to common everyday paper-cuts, little automations, and connectors.

Namely, I now have a script that uses my podcast OPML file and takes out relevant data into a yaml file. My site then can compute through liquid’s syntax site.data objects providing counts, links and latest update (helping to find dead feeds), similarly for my blogroll out of NetNewsWire (solves my needs, zero resale value).

I also have some Drafts actions that take words (like these) and put them into blog post formats (front matter and markdown) for various post types in Working Copy ready for committing and pushing to my blog. Drafts actions that get film data and push that up to my film archive.

The list goes on, an action that takes a list of items and costs, copied from a Notes app the better half uses and sums the total so I can transfer the funds (school trips and other ad-hoc costs). Another that takes a URL or YouTube username and returns available RSS feeds - as a keen consumer.

Scripts that sort lists, get the weather forecast, time zones, I’m sure I’m labouring the point now, but all these things make my life a little easier, can be run from my phone in most cases and just connect one data source into another for the consuming app, website, or person. Removing friction for me whilst there is little to no tangible value to anyone else - so not worth the time or effort in a team or business setting.

For those in teams, AI can be great at building the pipelines, connectors, and adapters that make time for the team to focus on the artisan work where the real value is.