Building Optional Income in the AI Era (Without Quitting Your Job)

I wrote a full product spec last Tuesday before my first coffee had gone cold. Not a stub. An actual feature spec, with context and edge cases and the reasoning behind the trade-offs. By the time Hazel started making noise from her room, it was done.

That used to take most of a morning.

AI has made building cheap in a way that's genuinely strange to sit with after spending a decade shipping things the hard way. MVPs over a weekend. UI in twenty minutes. You can get from idea to deployed faster than most teams used to align on the scope. Which means building is no longer where the constraint is, more or less.

The constraint is distribution.

For years, the work happened inside. Better flows, reduced churn, tighter funnels. The product was the bottleneck and product people were the ones holding the answer. Now anyone with a weekend and enough patience can ship something that works. The question stops being "can we build it?" and becomes something harder to answer: can you get someone's attention, and then actually do something with it once you have it?

People who can do both are building real options for themselves. Not in some vague early-retirement fantasy. In the concrete sense of not depending entirely on one employer for income, which is a different thing and worth being clear about.

Anyway, this is what's been sitting with me since Hazel arrived.

I'm not trying to quit my job. What I mean by optional income is reducing dependence - having something generating money outside the single paycheck, slowly building assets that compound, testing ideas without risking the salary covering the mortgage. The AI age makes this more reachable than it's been before. Lower cost to build means smaller bets, faster feedback. You don't need a business. You need a structured experiment, and then another one.

My current constraint is a feature, I think. (I keep saying this and then wondering if I'm just coping - but the discipline it's forced seems to be holding, so I'll keep saying it for now.)

Hazel is five months old. Emily is on parental leave until October. Sleep is unpredictable. If I protect the mornings right, I get two to three focused hours at the library, Monday to Friday. That's the whole budget. What that forces is ruthlessness about where you put those hours. You stop romanticising the idea that needs six months and a team. You pick the thing with the most signal per hour and get moving.

The first experiment I'm running is YouTube.

Not because YouTube is exciting or new, but because it solves two problems at once. Distribution and monetisation, in a format where the content compounds. A video you post this week can get found in eight months. There's a discovery engine built in, a clear path to revenue, and production costs have come down enough through AI that it's a real option for someone with a two-hour daily window and a knee injury keeping them off the trails.

The channel is called The Systems You Never See. Invisible infrastructure - water, power, logistics, the internet - and what happens when it breaks. I genuinely find this stuff interesting, which I think matters more than it sounds like it does.

The test: two to three videos a week, four to six weeks, then a real review. The goal at this stage isn't subscribers. It's signal - which formats hold attention, which topics travel, whether something can actually happen here or whether it quietly stalls. You can't build optional income on hope. You build it on data, and the data right now is four impressions since Saturday, which is nothing, which is exactly where you'd expect to be nine days in.

So I'm measuring click-through and retention. Not subscriber count, which is vanity at this stage. I've set a decision date six weeks out: double down, change direction, or kill it. Optional income experiments need you to stay emotionally detached from the outcome - easier to say than to do, honestly.

I'm not writing this up because I have the answers. This piece has been in my drafts for two weeks because I kept waiting until I had something more definitive to report, which was the wrong reason to wait.

The thing that keeps coming back to me is the perspective gap. AI can build things. It cannot make anyone care. That gap is where the real work is, and the people in tech who are going to have real options aren't necessarily the best builders. They're the ones figuring out distribution. Product managers especially are going to have to become something more like media operators than they're probably comfortable with. Builders who own their distribution will have leverage. Those who don't will still have jobs, but fewer choices than they think. I'm still working out which side of that I'm on, which is part of why I'm running this test rather than just theorising about it.

What success looks like at three months isn't revenue. It's one asset generating something, or close enough that the signal is clear. Better understanding of what actually travels. Less dependence on a single income track. Leverage builds slowly at first, then faster. That's the whole bet.

I'll keep sharing what works, what doesn't, what surprised me, what I turned off.

If it pays for the nappies, that's enough to start with.