Change and adaptation

This post isn’t my typ­i­cal nerd post. I’m in a con­tem­pla­tive frame of mind and just felt like writ­ing. 

Until the past year or so, if you asked me ​“what do you do,” I had an easy answer that I was hap­py to share. I’d tell you that I run my own dig­i­tal pro­duc­tion com­pa­ny with a great friend of mine. That com­pa­ny was called JMX2, which I co-found­ed and ran with my friend, Jon, for about 17 years. We found a way to do the work we want­ed to do on our terms and we had fun doing it. I often say that Jon and I got lucky. My hus­band would try to cor­rect me and say that hard work played a big role. Yes, we did work hard. But, a lot of peo­ple work hard and have a pas­sion for what they are doing. It doesn’t always work out. But we were lucky enough that it did for us. We were in the right place at the right time and our per­son­al inter­ests aligned with what the world was will­ing to pay us mon­ey to do.

In Decem­ber 2024, Jon and I offi­cial­ly closed JMX2. It’s now near­ing the end of 2026 which make our clos­ing date seem in the dis­tant past, but it’s still fair­ly fresh for me.

We closed our com­pa­ny because our work vol­ume had begun to slow down through­out 2024. One of our biggest clients spent the year pulling the con­sis­tent big projects we had been doing for them for over a decade back inside the com­pa­ny using inter­nal resources. They were mak­ing a big shift to focus on inte­grat­ing AI into their com­pa­ny and bud­gets for exter­nal ven­dors was an easy place to cut expens­es. Oth­er client work was also grow­ing scarcer. It just felt like it was the right time to close shop. Jon and I took our remain­ing client work and split it up. As the clock ran out on 2024, we com­plet­ed all the paper­work to offi­cial­ly close. 

It was only last month, on August 14, 2026, that we final­ly shut down the rem­nants of JMX2. We had main­tained our Google Work­space account since our offi­cial close. It made sense to us to keep our email and cal­en­dars going through 2025, our first year post-clo­sure. Then 2026 rolled around and not deal­ing with shut­ting the remain­ing bits down start­ed to nag at me. Over the sum­mer Jon and I got our old emails archived and cal­en­dars moved which let us safe­ly shut down the Google Work­space account.

That final piece of JMX2 shut­ting down made the clo­sure of the com­pa­ny more real to me. It has been a year and a half, so I’ve had time to adjust to it, but it still is a real mark for me. I guess it’s like final­ly get­ting a head­stone installed for some­one who has died. It’s the last step. That sounds more mor­bid than I intend­ed to make it.

I’m still work­ing at my com­put­er every day, but how do I answer the ques­tion, ​“what do you do?” I still find my day to day task list near­ly the same as before because what I’ve done for mon­ey was also my hob­by. Sure, I do have some remain­ing client work, but it doesn’t fill my day like it used to. I now make things for myself — art projects, soft­ware I’ve always want­ed and oth­er exper­i­ments. What do I do, though? I’m some­where between ​“I’m retired” and “ I work for myself full time.” Or at least that’s the clos­est descrip­tion I’ve found so far.

I’ve fall­en hard for AI cod­ing. I love it. I’m build­ing new projects all the time. When we had JMX2, it was the pre-AI era and we hand cod­ed every­thing by our­selves. We felt the pres­sure from clients to hire more staff, but we liked the build­ing process and the refine­ment process.

Some of the most fun projects we built at JMX2 were the ones where we were pri­vate­ly skep­ti­cal that we could pull off but knew, with enough effort, we could do it. That often end­ed up with the two of us work­ing every wak­ing hour, but it felt so reward­ing to ​“fin­ish the puz­zle” of each of those chal­leng­ing projects.

Now that I’m on my own, I find my own puz­zles to do. I have the cod­ing assis­tants that we nev­er hired while we had JMX2. I have Claude or Codex to go off and com­plete the tasks that I cre­ate. It’s been won­der­ful. It’s hard to think about giv­ing it up. Could I? Sure, I could try, but I doubt I will. I can fin­ish full projects so quick­ly and just start using them.

If you’ve read oth­er pieces on my blog, you’ll know some of these projects I’ve built. An upload serv­er, a secret shar­ing ser­vice, a cloning text to speech ser­vice, a deploy­ment check­ing ser­vice. Each one of these projects is some­thing I think in the before-AI times would be a pos­si­ble new SASS. I haven’t released any of them as paid ser­vices though because I feel that they’re all real­ly devel­op­er focused and any devel­op­er that I think could ben­e­fit from them now has at their hands the abil­i­ty, with AI, to just repro­duce their own ver­sions.

In the pre-AI era, I think sev­er­al of the projects I’ve built could have been viable prod­ucts or ser­vices to sell. The strange thing is that hav­ing AI cod­ing tools so read­i­ly avail­able has reduced the per­ceived val­ue of soft­ware as a whole. I think the things I’ve built are gen­uine­ly high-qual­i­ty, but I’m not sure most peo­ple would rec­og­nize — or be will­ing to pay for — the dif­fer­ence between soft­ware that has gone through three weeks of rig­or­ous UX refine­ment and test­ing and some­thing they can cre­ate in an hour or two by talk­ing to a chat­bot. That makes it hard to jus­ti­fy try­ing to sell them on the dif­fer­ence.

At the same time, I still think good soft­ware devel­op­ers are incred­i­bly valu­able. For exam­ple, yes­ter­day I had a friend show me his web­site that he was pay­ing some­one to build for him. He had spe­cif­ic ques­tions for me around aspects of a user check­out flow that I don’t think was being han­dled very well by the devel­op­er he had hired.

I talked to him about how I think about these types of prob­lems before I dive in and write any code. I think he would have ben­e­fit­ed from a much bet­ter devel­op­er than he’s cur­rent­ly work­ing with, but that will be up for him to decide when he gets the final prod­uct from the per­son he’s hired.

My assump­tion is that this devel­op­er will prob­a­bly be using AI behind the scenes, at least to help him or her cre­ate the site for my friend. But my impres­sion of what I saw as the end prod­uct left me with a lot of wor­ries about what’s going on. Since this friend was not ask­ing for me to give that kind of feed­back, plus we’d had that con­ver­sa­tion in the past before he made the deci­sion to go down his cur­rent path, so it did­n’t feel right to retread old ground.

But I want to cir­cle back to what that exchange made me feel about the val­ue of good soft­ware devel­op­ers.

Good soft­ware devel­op­ers who under­stand user expe­ri­ence and have empa­thy for the end user, com­bined with an under­stand­ing of how sys­tems work, are still incred­i­bly valu­able. I think the prob­lem is con­vinc­ing the peo­ple who need soft­ware, whether that’s a web­site or a new mobile app, that they need to either hire a skilled per­son or be will­ing to invest the time to teach them­selves the skills need­ed so they can talk to their chat­bots and get the end result that they want. But I don’t think most peo­ple are even aware of the skill set that they need.

As an exam­ple, I don’t feel like I have the skill set need­ed to do my own tax­es. Sure, I could fill out a form, but I’m com­plete­ly lost when it comes to fill­ing it out cor­rect­ly, and that’s why I hire some­one who is skilled at doing that chore.

How do I wrap up this long, ram­bling post? Let’s cir­cle back: I closed my com­pa­ny, and I’m not sure exact­ly what I do now, but, day to day, I still do many of the same things I used to do while I had my com­pa­ny, but now I use AI to help.

If I were start­ing that same com­pa­ny now, I would be fight­ing the bat­tle of still need­ing to con­vince peo­ple to pay me for what I do, and pay me well for it, while their per­cep­tions of the val­ue of what I am capa­ble of doing have been erod­ed by the exis­tence of AI.

Where does that leave me? Well, all this change is hap­pen­ing after I’ve had a good run through­out my career. If I had my choice, I’d still have JMX2 chug­ging along for a few more years, but that was­n’t meant to be. The AI shift hap­pened late enough in my career that, due to my ten­den­cy to plan for the worse and hope for the best, I’ll be okay. I just hope my younger friends will main­tain their foot­ing as the land­scape goes through the rapid upheavals we’re expe­ri­enc­ing. Good luck, friends.