Weekend Reflections #9 | Where the AI Meter Sits

[Views are my own]

Last weekend I bought a Bambu Lab 3D printer. My plan for this weekend was to keep learning, and I had decided to skip my weekend reflection. Nothing was forming, and I did not want to force it.

Then I got a notification. The relevant detail was not the printer itself. It was a pricing change in the software layer around it.

Bambu Lab’s MakerWorld had been offering AI-powered model generation with very little visible friction. In some cases, the visible cost appeared only at export. Generation itself felt almost free.

Then I read a community post announcing a change: some generative features would move from "consume credit upon export" to "consume credit upon generation." Monthly free credits would be replaced by limited trial uses. Direct purchase of MakerLab credits would also be introduced.

A new economy was opening inside a product ecosystem I had joined to print a small plastic bracket. At least, that is how it felt from a product perspective.

AI credits themselves are starting to become common. The more interesting product question is where the meter sits.

Charging at export prices usable output. Charging at generation prices exploration. That small change turns infrastructure cost into product psychology.

Much of the public AI conversation still follows one story: efficiency. Faster processes, fewer handoffs, lower operating costs. That story matters, especially inside companies. It shows up in earnings calls, productivity plans, automation programs, and operating-model discussions.

But there is a second story forming underneath it.

AI was already inside many consumer products. It recommended songs, ranked feeds, personalized offers, improved search, optimized ads, and helped companies sell more effectively.

But that was mostly AI behind the curtain.

The newer shift is different. AI is moving to the front of the product. It is becoming something users can see, request, consume, exhaust, renew, and pay for.

That is why credits matter. A credit is not just a pricing mechanic. It is a sign that AI has become a metered product capability.

Alongside the efficiency question, another question is becoming visible: what can we now offer, meter, and price that did not exist as a product capability before? That is a different question. And it leads somewhere different.

Adobe uses generative credits in Creative Cloud. Cricut uses credits for AI Project Designer in Design Space. Canva uses AI usage allowances, plan-based limits, and different tiers for standard and premium AI features.

Different products. Same shift: AI becomes visible, countable, limited, renewable, and monetizable.

Charging at export keeps experimentation feeling open and charges closer to usable output. Charging at generation moves cost earlier into the creative loop. That may be economically necessary, but it changes user psychology.

When export is priced, users can play. They can generate bad ideas, weird ideas, half-formed ideas. The meter appears when something is worth keeping.
When generation is priced, every prompt becomes a small purchase decision. The user starts asking whether the idea is worth spending on before the idea has had a chance to become good.

The product is no longer only managing compute cost. It is managing the user's willingness to explore.

This matters because the organizational logic is different.

When AI is a cost story, the ROI question is: how much do we save? The work happens inside the company, invisible to customers, measured in productivity, automation, process efficiency, and sometimes workforce impact.

When AI is a revenue story, the question is: what new value can we deliver and price? The work faces outward. It becomes part of the customer experience. Success looks different, and so does failure.

In these products, AI is not only a back-office productivity tool. It is becoming inventory on the product surface.

Many of the companies moving in this direction are not traditional software companies. They are hardware manufacturers, creative platforms, retailers, fitness companies, connected-device makers, and consumer brands that now have a reason to treat AI as part of the product experience itself.

We talk a lot about what AI does to the workforce inside software companies.

We talk less about what happens when the companies building physical things, selling subscriptions, and running consumer platforms start treating AI as a product layer rather than a process improvement.

There is a useful distinction between B2B and B2C here. In B2B, the AI moat often moves toward proprietary data, workflow context, governance, trust, and integration depth. In consumer products, the moat may move closer to taste: how well the product helps users imagine, choose, create, and refine before they can fully articulate what they want.

That second wave is quieter because it does not always look like disruption. Sometimes it looks like a button in an app, a credit counter, a trial limit, or a new export rule. Not in the news. In a notification about credits you did not expect to receive on a Thursday morning.

But those small product decisions reveal something larger. AI is no longer only changing how companies operate. It is changing what they can package, meter, and sell.

And when the meter moves from output to exploration, pricing stops being only a business-model decision. It becomes product design.

For product teams: do not ask only what AI costs to run. Ask where the meter belongs. Which part of the loop should feel free? Which part should feel scarce? And what behavior will that scarcity create?