Building got so fast the real work is idea generation

Picture a software engineer working on his startup in the year 2020.

Say he has about four new ideas on what to build per week. Implementing these new ideas varies, but let’s say it takes, on average, two days.

Within five workdays, our founder will have created about eight workdays of things to do (two days per idea, four ideas per workweek).

That’s a healthy cycle, and our founder will continue to build and come up with new stuff.

Zip forward to 2026. Agentic coding & loops are the norm.

Our founder still has four new ideas per week, but implementing them might only take two hours each. So within a day, a week's worth of ideas will have been implemented.

In 2020, implementation was the bottleneck. Today, creating new ideas is.

Flow used to be the job

Back in 2020, when I would implement a new feature, I’d be in a deep flow state, constantly thinking about how to solve a specific coding problem.

There wasn’t ever any waiting. I was the bottleneck.

New ideas would come in various ways: customer feedback, research, thinking, but I’d say the rate of these new ideas was and is pretty static.

It doesn’t matter when the next idea pops up - mid-coding, after shipping a new feature, or in the shower. What matters is the clock.

If four ideas a week each take a long time to implement, time itself will provide more things to work on.

You’re busy for months. In those months you still get more ideas at roughly that same weekly rate. So you never run out. You’re always slightly behind a backlog that grows while you build.

Well, not anymore.

Flipped ratio

This isn’t something that just impacts solo founders like myself.

I can imagine the same thing happening in any sort of software development-related business.

You have a Product Manager (PM) responsible for creating an idea, fleshing it out and prioritising it. Usually you’d have one of these and several developers to implement the ideas.

Say that ratio was 1:5. I can’t imagine it being at that level anymore. I don’t know where it is exactly, but I could imagine it being somewhere like 0.5:1, where in extreme cases, you have more PMs than developers (I’m not the first to notice this).

This then pushes the bottleneck upstream at an organisation level, just like it pushes up the problem for me personally.

If an entire engineering org starts running out of backlog, it’s easy to see why a solo founder - who is PM and the engineering team - hits the same wall.

I don’t currently have great insights into larger organisations, so all I can do is focus on myself and the view I have on solo founders.

Backlog implemented by lunchtime

A typical day for me will start with reviewing customer feedback. This results in a new feature or a bug to be fixed. Since this is my highest priority, I do this first thing. But with automation, this doesn’t usually take that much actual doing.

So by midday, these high-priority things are off the table. This process will happen a couple of times a week.

Come afternoon, there is not much backlog left; everything has been implemented.

In 2020, I didn’t have to worry about deliberate idea generation. The fact that it took so long to implement meant it was never a problem. The process of sitting down to find new ideas, for me, is undefined.

I never needed it, yet now I feel like I need to spend more time on deliberate idea generation.

Where ideas actually come from

Ideas can come from a lot of places, such as exploring other products, talking to customers, or idling, as in not being distracted

For me, the customer channel is already maxed. Every second I get, I’m using. There just isn’t enough volume yet to keep my backlog filled.

Maybe I’m only seeing a small part of the picture because I don’t have product-market fit with a ton of customers leading the way and telling me what to build.

In general, I feel like ideas can’t be forced. If you push and think about these ideas too hard, they often don’t come.

When you idle and are not distracted, however, things seem to pop up.

But that’s not the natural path for someone who is used to grinding through a backlog.

On top of that, stepping away to be undistracted feels like not working.

The guilt of not looking busy

Personally, if I don’t spend my workday in front of the computer pushing, building, doing, I feel guilty of not having given it a good go.

But sometimes that might be what is required. Sometimes to think, process and get a view from a higher vantage point, one needs to step away and actually idle.

By idle, I don’t mean watching YouTube or scrolling Twitter. I think what is needed is true downtime away from the computer and phone.

The new bottleneck might look like not working: a walk, reading a book, or taking the kids to the beach.

When I think about it, it makes sense. The cost of execution drops so much that the only way to differentiate is that high-value work of seeing the big picture.