Just-in-time Personalized Software
We now live in a world of zero-cost software, and it’s hard to overstate how fundamental of a shift that is at every level.
The market is starting to realize this, selling off a bunch this week after Anthropic rendered a category of applications effectively useless with its legal tool (AI is Killing SaaS). There has been much chatter about the AI bubble fueled by CapEx investments, but in comparison much less about the long-term effects on companies like Figma & ServiceNow, down 80% and 45% in the last year. What’s holding SaaS up is inertia and distribution (and, let’s be honest - security, compliance, exec trust, etc. aren’t going away in the short-term and will keep the status quo going for awhile).
Work is starting to realize this. Whether we like it our not, every business must assume our competition (both currently existing and yet to come) will be fully equipped with AI tooling both augmenting and authoring code (and eventually everything else). Maybe there is some counter-positioning strategy advantage that comes from being mostly hand-written code, but currently the arms race is fully engaged. The feeling is existential - we must adopt or die. At an individual level this changes the workplace dynamics significantly, but both the good and the bad news is that almost all developers are aware. The good is that that everyone is undergoing this massive change at the same time and we’re all in this predicament together, so all starting from the same piont; the “bad” is that rather than being an opt-in accelerator (traditionally, this might be spending free time building projects, taking courses, etc.), it is functionally a requirement just to keep pace with peers, and all of your peers. Bad might mean “painful” here if you’ve excelled in the previous way the software workplace worked, or just don’t like AI; there is almost a career-switch level of change that might be required for some folks.
Individuals may be starting to realize this too. Clawdbot / Moltbot / OpenClaw went viral this past weekend. Most of the hype was around Moltbook, and it seems like that might’ve been faked to some degree, but the main takeaway here will be the mindset shift for folks who watched or participated; it went viral for how believable it was, because the leap from where we are today to a world of OpenClaw agents is just a small step and probably inevitable to some degree.
What’s most exciting to me about this shift is also about the impact at the individual level. It’s the potential for just-in-time personalized software: if you have an idea, just build it. No reason to wait to build it either - just build it as soon as you have the idea. Leveraging AI coding tools paired with AI interfaces allows for compressed creativity, where the main limits are a properly setup automated build harness, a bunch of tokens, and coming up with ideas. The constraint of time no longer exists.
There are so many small tools and/or utilities that could be built that may be helpful:
- Weekly meal generator that will fit to your dietary goals (read: a diet after the holiday season) that figures out your aggregated shopping list and emails it to you
- Auto-syncer of expenses into your budget spreadsheet + tools (I do this by hand now, and it’s a bit of a ritual, but also a bit of a time suck - and, I have tried Monarch & previously used Mint, but I like my spreadsheets better)
- Custom TODO app that’s tailored to your level of engagement and detail (I’ve used so many and always come away with disappointed; my go-to is now an unorganized iPhone note & a wall of post-its)
- Automated Cocktail recommendations based on what bottles I have
I mean, there’s nothing revoluntionary there. A few of those are probably technical interview questions for a mid-level engineering job. They definitely could’ve been built at any time over the last 10 years.
Software has always enabled us to build personalized tools. It’s just that before, the ROI was just not high enough for the time & maintenance. And, who knows if by the time it’s finally built, I’d really even care that much about it anymore (my diets typically last all of two weeks!).
The scarce resource is no longer time, because that’s outsourced to Claude or Codex or something else. Paired with a Ralph Wiggum approach or a Gas Town (I am only Stage 6, so not quite to Gas Town levels yet), you can kick something off before bed and wake up to it baked (maybe half-baked) in the morning, and start to actually use it pretty shortly after. I can easily envision in a year’s time running the rote or high activation energy parts of my life with a bunch of small automated utilities. Building on top of that, there are many areas of enrichment that I’d like to do, but can’t due to a lack of time (and poor time management). It’s pretty exciting to think about what happens when lack of time is no longer a reason not to do something.
Personally, my list of project ideas has always been way longer than my ability to complete them; now, that’s no the case.