huhubWhat becomes scarce
If you own a business, AI has made an old question more urgent: what is worth building? You can see more possibilities, commission more experiments, and reproduce more of what already exists. You still have to decide where to commit your resources, what must survive every iteration, and why someone will choose the result.
The change is real. In May, METR reported strong performance on an early benchmark of 25 software replication tasks, using black-box access to existing command-line programs. These were bounded tasks with unusually clear feedback. Reproducing selected software behaviour is becoming more accessible; reproducing a complete business remains a much larger claim. METR, May 2026.
A September revision of an NBER working paper, covering more than 500,000 developers, estimates substantially more coding activity and 30% more releases with the latest generation of tools. Across four software marketplaces, more applications appeared without an increase in aggregate usage. This observational study captures a useful distinction: creating more possibilities does not automatically give people more reasons to use them. Writing Code vs. Shipping Code.
Money is moving into AI while the outcome remains uncertain. In a survey published in October covering 10,000 businesses across 32 countries, 65% planned to increase AI spending; 28% reported broad or strong returns. These are businesses' own assessments, not audited results. Dun & Bradstreet, October 2026.
A good product still takes considerable time, judgement, and care. What AI changes for my practice is how much of that work a smaller team can carry, from intent to operation. I want fewer gaps between decisions and responsibility for the whole experience. That responsibility is a working commitment we choose, not an automatic effect of using AI.
I do not want to compete with AI on the volume of analysis. My attention does not scale the way computation does. My advantage begins with what I care about: desire, intuition, taste, lived experience, and the courage to choose. I use rational checks and external evidence to test implementation and discover where my judgement fails.
A dream becomes a direction
This is where a dream becomes a strategic asset for me. It gives my work a direction I personally care about, and a reason to stay with difficult decisions. Each product becomes a way of discovering how to make that direction real.
I call the principle that connects that dream to decisions a metaidea. It can generate several products and help me choose what to leave out. When a technology changes or one product fails, it should still help me find the next meaningful form. Its value becomes visible in choices, including the features I reject and the experiences I preserve.
For huhub, that direction is to expand people's ability to learn, feel, and create something of their own, in contact with other people and the world. The dream provides direction and motivation. The metaidea guides selection. Products make it tangible. Lived experience and the market test whether those choices hold up.
The only path I see for my own work is to keep developing this source of connected products. Each experience can teach the next one something. A family becomes valuable when those connections help people, rather than merely making the catalogue larger. Its principles, interfaces, and features can all be copied. Its strength has to keep earning a place in actual use.
The smallest complete experience
My unit of work is the minimum viable experience, or MVE: the smallest complete experience that carries the intention. I am its first demanding user. I want to feel satisfaction, delight, and the desire to return. Scope can shrink while the heart of the experience remains intact. My own response establishes an authorial standard; other people's experience and willingness to pay require separate checks. A video showing a game made in ten hours can show an impressive prototype. It does not establish whether the experience holds together in real use. The product I put my name on must be something I am proud to show and stand behind. That takes time.
AI has expanded how much of this process I can carry. I take responsibility from intent through product decisions, architecture, implementation, and acceptance, including the agreed arrangements for operation or handover. The people and tools can change. Responsibility for the result I promise stays with me.
Where the idea takes form
One part of my workshop explores learning through play. The intention is to help people acquire an ability through their own actions and choices. The product is still private, so its design and release details are not part of this public story.
That emphasis on agency has an audience, although a survey cannot validate this product. In a UK study of 40,543 teenagers, 62.8% agreed that AI should help them think rather than think for them. National Literacy Trust, August 2026.
My music and instrument building give me another way to judge an experience: through timing, touch, attention, and expression. Facet turns polygon geometry into sound and a visible trajectory. NFZ, an evolving musical system for shared physical settings, is designed around a complete composed session. Sensors must earn their influence by improving that experience. Full-session evaluation remains open; this is an artistic product, with no claim of medical benefit.
Trust also lives in concrete decisions. In a survey of 11,000 online consumers across 12 countries, 56% valued knowing when AI was used more than getting the fastest result. That is a stated preference, not a purchase. It still gives us a clear design question: can people understand what the system is doing and retain a meaningful choice? Ping Identity, October 2026.