Some reflections after a year with AI

TL;DR: Nothing mind-blowing here. Just some musing on how I think the last year of AI/LLM agents has been quite okay for us.

At Snabble, we've got a strong culture of doing what makes sense. It feels sober, reasonable, pragmatic. We're careful about how we spend our innovation points.

So we obviously didn't dive head-first into the AI hype.

All this stuff about token-maxxing, PRs full of AI slop, and AI burnout that I'm hearing about in the industry, it all seems quite foreign to me (maybe because it's mostly occurring in foreign companies and startups).

We've been easing into AI-assisted development over the last year or so.

Some colleagues adopted early, others later on. By now, everyone has a Copilot or Claude subscription with a medium-sized budget. People use it where it makes sense for them.

There's no pressure from management. We do a monthly get-together to exchange tips on what new models work well for which purpose, what tooling are people trying out, show & tell stuff.

How it helps me individually

Does the agent help me write good code? Yeah, absolutely. It makes mistakes, I guide it and some times I just do things myself. We make a good team. I don't feel like it's taking the fun out of programming.

Have I been swamped with more code to review? Not really.

Does the agent help me doing code reviews? Sometimes with technical stuff, but it is missing too much context.

It does help me by kicking things off. That's actually one of my favorite things about it: if I find myself procrastinating some task, asking the agent is a great way for me to get in the zone. Like an energetic pairing partner ready to go, it's got great "attitude" :)

And it's great at completing some boring, simple chores for me.

I know it's speeding some things up, mostly some new frontend spikes and prototypes. Hackathons are a lot more fun for sure!

So what about quality?

At the end of the day, we (my team) need to sign off on what goes into production. So we review and work on AI-generated code just as thoroughly as we do on hand-written code.

Sure we can vibe-code a new feature and just ship it quickly. In those cases we make it clear that such shipments come with a strong disclaimer:

it may not work as expected,
quality may be low, and
it'll require real investment to make it a fully fledged part of our product.

I'm sure that over time, those vibe-coded shipments will get better. Our code will be less and less hand-written. And I think that's okay. Using a daft metaphor: Like with airline pilots, the autopilot will take over more and more work, while we're still around to decide the destination, keep things in check and be ready to take over if the machine fails.

Comments

Git Stash Blooper (Could not restore untracked files from stash)

The other day I accidentally did a git stash -a , which means it stashes *everything*, including ignored output files (target, build, classes, etc). Ooooops.. What I meant to do was git stash -u , meaning stash modifications plus untracked new files. Anyhows, I ended up with a big fat stash I couldn't get back out. Each time I tried, I got something like this: .../target/temp/dozer.jar already exists, no checkout .../target/temp/core.jar already exists, no checkout .../target/temp/joda-time.jar already exists, no checkout .../target/foo.war already exists, no checkout Could not restore untracked files from stash No matter how I tried checking out different revisions (like the one where I actually made the stash), or using --force, I got the same error. Now these were one of those "keep cool for a second, there's a git way to fix this"situation. I figured: A stash is basically a commit. If we look at my recent commits using git log --graph --...

Git-SVN Mirror without the annoying update-ref

This post is part of a series on Git and Subversion . To see all the related posts, screencasts and other resources, please click here . So no sooner than I had done my git-svn presentation at JavaZone , I got word of a slightly different Git-SVN mirror setup that makes it a bit easier to work with: In short, my old recipe includes an annoying git update-ref step to keep the git-svn remote reference up to date with the central bare git repo. This new recipe avoids this, so we can simply use git svn dcommit directly. So, longer version, with the details. My original recipe is laid out in five steps: Clone a fresh Git repo from Subversion. This will be our fetching repo. Set up a bare repo. Configure pushing from the fetching repo to bare repo In the shoes of a developer, clone the repo Set up an SVN remote in the developer's repo In the new approach, we redefine those last two steps: (See the original post for how to do the fir...

Syncing your Git repo with the Subversion repo

This post is part of a series on Git and Subversion . To see all the related posts, screencasts and other resources, please click here . This is a follow-up to the previous post on how to live with Git and Suversion in parallel . It's really trivial stuff for any experienced user, but was worth noting down somewhere for my own sake. I've cloned a project called "fudge" from Subversion: >git svn clone https://scm.mycompany.com/svn/fudge So my local repo has the correct svn-remote configuration and all that to go with it inside fudge/.git/svn . This is done automatically when you svn-clone. The bad thing is that all my Git-mates at work have to wait for me doing svn-rebase before they can pull the latest code from Subversion from my repo into theirs. I want to get rid of this responsibility. I'll put something similar to my svn-rebasing repository on a server, have it run svn-rebase regularly, and push the changes to a centralized git repository, ...

Git-SVN Mirror for multiple branches

This post is part of a series on Git and Subversion . To see all the related posts, screencasts and other resources, please click here . This extends the posts where I explained how to set up a git-svn mirror for a single directory. NOTE: If you just want to use Git against a SVN repo on your own, stop reading ,now, and stick to the git-svn basics. However, if you want a setup where you can share a Git repository with colleagues and friends while still interfacing with Subversion, keep reading. I'll show how to set up a git-svn mirror for a standard Subversion project with trunk , branches and tags . It's a bit like the single directory mirror, but in order to keep all branches in sync, it's a bit more fiddling. The good part is that this setup enables us to cherry-pick commits from one branch to the other. This is slightly smoother than using svn merge . First of all, let's repeat how our Subversion and Git-repositories look physically (roughly the sa...

The Dream of a Bi-directional Git-SVN mirror

This post is part of a series on Git and Subversion . To see all the related posts, screencasts and other resources, please click here . I just got an email asking me how one can set up a bi-directional Git-SVN mirror. It ended up being quite a long answer, so I'll post it here for the benefit of other Git-SVN readers with the same idea. As you may know, I'm a proponent of my own Git-SVN setup . I remember trying to go down the path of a bi-directional repository, but always ran into problems. Here is how it could work: However nice this would be to have, it can be very hard to achieve in practice: Git-svn requires working in a non-bare repository, so pushing to it is by default refused. You can work around this by doing this in the target sync repo: git config receive.denyCurrentBranch ignore You also have to automatically perform a git reset --hard in the syncing repo after each push (by some git hook?), because the work-dir is c...