> ## Content Index
> Fetch the complete content index at: https://robweidner.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Don't Let Tomorrow's Problem Choose Today's Tool
- URL: https://robweidner.com/tomorrows-problem-todays-tool/
- Published: 2026-11-13T17:24:32.000Z
- Updated: 2026-11-13T17:24:32.000Z
- Description: I see a lot of teams pick tools for problems they might have in a year, while the problem in front of them waits. AI made building faster. Change is still hard on the people who have to live with it, and in my experience that costs more than living with a tool that doesn't do everything you want.
- Author: Rob Weidner, PMP

****The short version:** I see a lot of teams pick tools for problems they might have in a year, while the problem in front of them waits. AI made building faster. Change is still hard on the people who have to live with it, and in my experience that costs more than living with a tool that doesn't do everything you want.

I watched a team almost pass on the tool that could fix the problem in front of them.

It would have worked. What worried them was a limit they might hit in about a year. The tool caps how many records it can hold, and they could see a future where the business ran into that cap. Fair concern. Moving everyone onto a tool, then moving them again a year later, costs real time and real money.

So the conversation drifted away from the problem they had and toward the problem they didn't.

That drift is everywhere right now. AI is making it more expensive.

## Just because you can doesn't mean you should

The pitch goes like this. AI made building fast and easy. So skip the middle tool. Build what the bigger, future version of the business needs.

Half of that is true. Building got faster. I do it every week.

The other half smuggles in an assumption: the cost of a tool decision is the cost of building it.

It never was. Economists have studied switching costs [since the 1980s](https://doi.org/10.2307/1885068?ref=robweidner.com). [In 2003](https://doi.org/10.1177/0092070302250897?ref=robweidner.com), three marketing researchers, Burnham, Frels, and Mahajan, sorted them into three buckets:

1. **Procedural:** The time and effort of changing
2. **Financial:** The money
3. **Relational:** The discomfort of breaking ties with people and brands you know

Paraphrased from Burnham, Frels and Mahajan, *Journal of the Academy of Marketing Science*, 2003.

They were writing about consumers, but watch a company change systems and you'll see all three. Someone moves and checks the data. Everyone relearns the routine. The business keeps running through all of it.

AI didn't repeal any of that.

[DORA](https://dora.dev/insights/balancing-ai-tensions/?ref=robweidner.com), Google Cloud's software research program, described one current example: developers save time generating code, then spend a chunk of it verifying what the machine wrote. A faster author leaves the reviewer with more work. Some of the work didn't go away. It moved to the next person.

![A robot produces pages while a person reviews a large stack of documents.](https://storage.ghost.io/c/9d/fd/9dfd4f1e-033b-45ef-89e9-0447dd89a095/content/images/2026/11/section_01_generation_vs_verification.jpg)

Writing it got faster. Checking it still takes a person.

Building got easier. Absorbing the change didn't. Your team's ability to learn, check, and trust a new system is the same budget it was last year. Spend it on a problem you have.

### The people are the expensive part

People forget the human side of this. The people are usually the hardest part of a switch. Everyone has to learn a new routine and give up the shortcuts they built around the old tool. That's change management, and in my experience it's more expensive and harder to get over than living with a tool that doesn't do everything you want. Compromising or slowing down costs you a feature or a few weeks. A rushed switch can cost you the team's trust in the next change.

### The context-switching tax

There's a context-switching tax, too. During a transition, people often have to keep the old system running while learning the new one. Check a number here, update it there, stop to remember which system has the latest answer. Every jump pulls attention away from the work itself. That tax can outlast the migration if the new tool becomes one more place to check instead of replacing one. Count the attention it takes to use the tool as well as the hours to build it.

## Is it actually a problem?

![A detective examines a dripping faucet with a magnifying glass.](https://storage.ghost.io/c/9d/fd/9dfd4f1e-033b-45ef-89e9-0447dd89a095/content/images/2026/11/section_02_is_it_a_problem.jpg)

First question: is it actually a problem?

Before anyone builds anything, I ask the boring questions:

- What problem are you trying to solve?
- What's the [job to be done](https://hbr.org/2016/09/know-your-customers-jobs-to-be-done?ref=robweidner.com) here, meaning what does someone need to get done when they open this thing?
- Can someone already do that job with the tools you have, and they just don't know how?
- And is it a problem at all? Or did someone just come up with a solution for the sake of it?

That last one comes up more than you'd think. AI hands people tools and systems they never dreamed of before. Someone takes the reins and jumps five steps ahead because they could. That's when switching costs get so expensive. The team gets moved onto something new, then moved again, for a problem nobody had.

Is it empowering? Of course. Just be careful.

Founders already know this one. Paul Graham told them to [do things that don't scale](https://paulgraham.com/ds.html?ref=robweidner.com), meaning do the work by hand before you build for scale. [37signals](https://37signals.com/?ref=robweidner.com) says [hire when it hurts](https://37signals.com/podcast/hire-when-it-hurts-rework/?ref=robweidner.com). Tools work the same way. Wait until the pain is real, and fix the pain you have.

## Sometimes you just need a button

One thing never makes the build-vs-buy spreadsheet.

I've made changes inside an existing tool that took less than five minutes. Add a button next to the other buttons. Have it do one specific thing. Done.

That speed came from the tool. The placement already existed. The styling already existed. The tool had already made most of the decisions, so I didn't have to.

> Constraints are a feature. They keep a small change small.

A tight budget works the same way. [Mike Michalowicz](https://mikemichalowicz.com/?ref=robweidner.com), who wrote [Profit First](https://mikemichalowicz.com/profit-first/?ref=robweidner.com), told a story on [Paper Napkin Wisdom](https://www.papernapkinwisdom.com/post/episode-139-sales-profit-expenses-mike-michalowicz-entrepreneur-author?ref=robweidner.com) about a leather maker he co-owned. That industry runs on heavy presses that cost tens of thousands of dollars apiece. Once the company started setting its profit aside first, there wasn't money for one. So the team went to Home Depot, bought parts and started experimenting. Mike hinted that home microwaves were involved. They ended up with a mechanism that costs about $200 and makes leather soft enough to mold. He said margins and productivity both went up. If they'd had the money for the next press, nobody would have gone looking.

I run my own business on Profit First too, and I [wrote about how my setup works](https://robweidner.com/profit-first-changed-my-financial-life/). Every deposit gets split into buckets before I can touch it. Earlier this year, when I was weighing a big annual subscription for a tool, I modeled it against my real numbers instead of guessing.

![A single red button beside an open panel of complex wiring.](https://storage.ghost.io/c/9d/fd/9dfd4f1e-033b-45ef-89e9-0447dd89a095/content/images/2026/11/section_03_just_a_button.jpg)

Sometimes all you need is one more button.

AI can add the button too. But give it the reins and you have to spell out what it shouldn't do:

- Don't redesign the page.
- Don't invent a new pattern.
- Don't turn a small request into a project.

More freedom means more guardrails, and every guardrail is something you now specify, check, and maintain.

I wrote the other side of this in my earlier piece on [no-code versus custom](https://robweidner.com/code-vs-no-code-2026/). Leave a platform and you inherit what it was quietly doing for you. Security, backups, maintenance. Even with managed services, someone owns how it all fits together. That someone is you.

## When fast is the right call

Sometimes speed is exactly right.

If you have the horsepower and the fuel to go down a road, turn around, and try another, you can beat the driver who has to choose carefully at every junction. AI makes those experiments much faster to run. Run them.

![Aerial view of a road making a sharp U-shaped turn.](https://storage.ghost.io/c/9d/fd/9dfd4f1e-033b-45ef-89e9-0447dd89a095/content/images/2026/11/section_04_reversible_route.jpg)

Pick a route you can turn around on.

But check who's on the ride:

- Who verifies the work?
- Who can reverse a change?
- Can the team absorb another iteration while doing their actual jobs?

A team with slack can choose speed. A team already stretched by change can choose steady. Both can be right. The mistake is assuming that because the implementation can move fast, everyone affected by it can move at the same pace.

---

## Steal this: five questions before the build plan

> The hard part is exercising restraint.

These five questions make it easier.

Buy, build, partner, or stay and improve what you have. Before you pick, get agreement on five things:

1. **The problem today.** What's slowing the business down or creating risk? Is it a real problem, or a solution someone wants to try? What improves the week it's solved?
2. **The future constraint.** What might become a problem, how credible is the forecast, and what evidence would make it real? Set a review date while there's still time to respond.
3. **The full responsibility.** Who builds, verifies, secures, supports, and maintains it? Who owns the next move?
4. **The capacity for change.** How much disruption can the team absorb? Who keeps today's work moving while you experiment? Will this remove a place people have to check, or add another?
5. **The next reversible step.** What can you test without committing the whole team, and what tells you to continue, stop, or turn?

**My default:** solve the problems the business has today. Keep tomorrow's visible. Write down the trigger that turns tomorrow's problem into today's, and it will rarely surprise you.

One caveat. If the constraint is real and close, that's a different decision. A limit arriving before you could responsibly migrate deserves action now. A limit you might reach in a year deserves a plan and a review date. Save the rebuild for when it's close.

And when you do reach the limit you planned for, that usually means something went right. You grew into it. The tool that got you there did its job.

---

**Which future constraint is steering a tool decision in your business?** And what would tell you it's become a problem worth solving today?

If the five questions surface a problem you can't quite name, that's the work I do as a [fractional CTO](https://robweidner.com/cto/). Reply with the constraint you're staring at.

Rob