🌳 Full postNovember 16, 20265 min chase

Unlimited Optionality (and a Serious Squirrel Problem)

AI hasn't cured my shiny object syndrome. It might actually be helping me manage it.

#AI#Foundation Roster#Development

If you couldn’t tell from the website, I have a squirrel problem. Not the furry kind digging holes in my yard. The kind where someone mentions an interesting idea, a new technology catches my attention, or I run into a problem that surely could be solved with a little software.

And then my brain says - COOL! LET’S CHASE THAT! What if I built that? How would it work? Could I make it better? What if I approached it from a completely different direction?

Soon, I’ve mentally designed half the product, identified three potential markets, and started thinking about the architecture… meanwhile, there’s an actual business I’m supposed to be running.

Squirrel!

My dog staring through the screen door at a squirrel on the hill, circled in red and labeled SQUIRREL

Curiosity used to be expensive

I’ve been building software for a loooong time. Having ideas is easy. Figuring out whether they’re worth pursuing is considerably harder. Historically, exploring an idea meant making a fairly significant investment.

Maybe I needed to learn an unfamiliar technology. Maybe I needed a specialist. Maybe I had to spend a week building something just to discover that the whole premise was flawed.

And because I’d already invested that time, there was always the temptation to keep going. Well, I’ve come this far… So much sunk cost fallacy.

That’s how a passing curiosity becomes a side project (Pocket Lanes (cough cough)). And how a side project becomes something you now have to maintain… Not exactly a great use of limited resources.

Enter AI. And more squirrels.

You’d think giving AI to someone with shiny object syndrome would be a terrible idea: “Hey! Build things for FREE!! You WANNA do it, right?!?!? It’s so EASY!!”

What could possibly go wrong?

I mean, yeah… at first, I did build all the things. But something else did, too. AI gave me a way to explore ideas without necessarily committing to them. I can take that half-formed (let’s be honest… half-baked) thought, get it out of my head, and start pulling on the threads.

What would it take to build? What are the alternatives? What am I overlooking? What’s the ugly part nobody mentions in the marketing materials?

Sometimes I’ll build a quick prototype (hello, “SCPDrop”). Sometimes it’s just an extended conversation exploring the possibilities.

And for the most part, I can walk away from it. That idea isn’t worth pursuing.

The squirrel has been investigated. No new project required.

A real example: Text messaging

At Foundation Roster, we build student management software for tutoring centers. Naturally, communication between centers and parents is an important part of the experience.

We already have SMS functionality, but I’ve spent a ridiculous amount of time exploring what it should look like.

  • Should messaging live entirely inside Foundation Roster?
  • Should we own the phone numbers or let customers bring their own messaging providers?
  • Should I totally rip out Twilio?
  • What happens when a customer already uses Comcast for their business phone service?
  • Could we use a provider like Clerk Chat to add texting without disrupting their calls?
  • Should centers use their existing business phone numbers?
  • Should Foundation Roster manage the conversations, or just integrate with a service that already does it well?

And who handles billing, compliance, provisioning, and support?

Every answer introduces another set of tradeoffs.

I’ve explored Twilio, hosted SMS providers, customer-owned accounts, embedded inboxes, webhooks, and more architectural variations than I’d care to count.

And we’re still exploring (“we” being myself, ChatGPT, Claude, Grokbot, and Gemini too, sometimes).

Only two years ago, investigating all those possibilities would’ve taken an insane amount of time, probably involved multiple vendors, and potentially required building several proofs of concept.

Today, I can work through much of that thinking with AI before writing a line of production code. Does it magically make the decision for me? Nope. It certainly doesn’t eliminate the need to verify technical claims or talk to vendors. But it helps me understand the landscape, identify tradeoffs, and figure out which questions are actually worth answering. So enormously valuable.

Unlimited optionality

This morning I said that AI gives me “unlimited optionality” as a technical founder.

I borrowed the language from options theory, where the value of an option comes partly from having the right, but not the obligation, to pursue something… That’s what this feels like.

  • I can explore an idea without committing to building it.
  • I can investigate five different approaches without needing five different teams.
  • I can discover that something is technically possible but economically ridiculous—and move on.

Sure, “unlimited” is a bit of an exaggeration. My time, attention, and resources are still very much finite. But the number of possibilities I can meaningfully explore has expanded dramatically.

And perhaps more importantly, the cost of deciding not to pursue something has dropped.

The catch

There’s still a trap here.

Just because AI makes it easier to build something doesn’t mean anyone needs it. And just because I can explore twenty ideas this afternoon doesn’t mean I should (but hey, it’s fun, right?).

AI can help me evaluate technical feasibility, compare approaches, and challenge assumptions. It can’t replace talking to customers, understanding their problems, or making the difficult decisions about where to invest. That’s still my job.

There’s also a very real danger of getting so caught up exploring possibilities that exploration itself becomes the distraction. Apparently, even squirrels can have squirrels.

Still chasing squirrels

Curiosity is a big part of why I became a software developer in the first place. It’s why I enjoy building things, solving problems, and figuring out how seemingly unrelated technologies might fit together. AI has lowered the cost of satisfying that curiosity.

I can follow an idea far enough to understand its potential without necessarily turning it into a commitment. Sometimes that means discovering something worth building… sometimes it means discovering something I definitely shouldn’t build.

Both are wins.

AI hasn’t cured my shiny object syndrome. It’s just made chasing squirrels a whole lot cheaper.

Nearby distractions

SCP Drop UtilityNov 15, 2026Keeping Foundation Roster Help Articles in Step with the ProductOct 25, 2026Transcribe Audio with Microsoft WordOct 16, 2026