For decades I felt that part of my job for my employer was to answer the question: “If X, then Y?”
There are countless examples from my career that I could provide, but here are a few that should ring a bell for anyone over cough 30 cough.
- If broadband internet becomes ubiquitous, more commerce will move online.
- If bandwidth costs go to near zero, much of the traffic will be video.
- If everyone owns a mobile phone, their first experience with a website will be mobile.
- If browsers become capable enough, more software will become web apps.
- If more commerce and content consumption moves online, ad revenue will become an enormous business.
You get the idea.
None of the above felt certain before we saw them happen. Using the internet used to be painfully slow. Distributing a 480p video online was very expensive. Software used to come in boxes and was updated once a year.
Part of my job was to see these changes coming, think through what they would mean, and help decide what we should do about them.
Every part of how we do what we do has changed. And it feels like the pace keeps increasing. The senses I had honed over the years to help me make decisions on what was coming next feel dulled.
Here’s one example. Eighteen months ago, I thought that being a software developer would mean that AI would replace me and I’d be left with little to do. And now, I know that I have more to do than ever before, and the tools to do all of that work.
We in the tech industry have been very used to change. In fact, the pace of change has burned out many of my peers over the years. But what we’re seeing now feels several steps beyond what we were used to.
In July I wrote about this sprint. Well, here we are just over three months later and it is going far faster than it was in July.
It isn’t just how we do things that is changing but also why. The incentives for software developers hoping to create products that they sell are shifting, in my opinion.
It used to be you’d try to sell a human on the software’s ease of use. A UI they understand and may even really like. You’d spend a lot of time deciding what to include, what to leave out, and how to make the things you did include easy to find.
Making software simple was the point.
But now I believe that making software ready for agents, and fully capable, will increasingly win out. This is part of what I was getting at in April when I wrote about the agentweb.
If the primary way a customer interacts with your software is through an agent, then having tons and tons of features doesn’t necessarily make it harder for that customer to use. They don’t have to learn where every feature lives, what each menu contains, or which sequence of buttons gets them what they need.
I’ve already seen this in my own use of Signboard. In March I wrote about brainstorming a project with an agent and then having it create the board, cards, due dates, subtasks, and labels. All of those capabilities are useful. But I don’t need to work through each of them myself to get the result.
That changes how I think about what belongs in an app.
Developers can add capabilities that might previously have seemed too complicated to expose in the interface. The “human views” for the apps can be simple, pretty, readable. Behind the scenes, the data, MCP tools, APIs, etc. can support much more than a person would ever want to navigate manually.
That still has to work well. The agent needs to understand what the software can do and use it reliably. But the customer doesn’t need to learn every part of it.
Making software simple may increasingly mean making it possible to do more without asking the person using it to learn more.
And yes, this is another “if X, then Y.” I’m still trying to work out what comes next. I’m just less certain that the instincts that served me for decades will give me the right answer. I’m relearning how to predict these things. And, certainly, the time horizons are shorter than ever.
For years, those instincts helped me decide what to leave out of software. Now I’m wondering how much more it should be able to do.
When the why changes… so does everything else.