The shift from analyzing technology to shaping it and what that means for anyone who wants to stay relevant in the age of artificial intelligence.
But there’s a difference between understanding technology and applying it to real problems.
The Shift from Observer to Creator
For the past year, I’ve been the guy analyzing AI trends, explaining SLMs, and teaching prompt techniques. It’s been valuable ! I’ve connected with many of you navigating this new landscape.
But there’s a problem with being the “AI guy” who only writes about AI:
It’s easy to fall into the trap of only consuming the technology and not shaping it.
And in a world where AI is moving this fast, consumption isn’t enough. You need to be creating.
That’s why I’m shifting from thought leadership to hands-on exploration.
What I’m Actually Exploring
Remember mem.ry? The AI-powered memory assistant I’ve written about, the one inspired by my dad’s Parkinson’s and memory challenges?
I’m not just writing about it anymore. I’m prototyping it.
But here’s the thing, I’m not trying to solve everything at once. I’m exploring one specific, meaningful piece:
How AI can truly personalize memory assistance.
Why Start with Family Recognition?
Because this is where memory support gets personal.
When someone with memory loss looks at a photo and asks, “Who is this person?” — that’s not just a recognition problem. It’s a context problem.
- Generic AI might say:
“This appears to be a young woman in her twenties.” - A personalized memory assistant should say:
“This is Emily, your granddaughter. She loves horses and usually visits on Sundays. Last time she was here, you two talked about her new job at the veterinary clinic.”
That’s the difference between information and memory.
And that difference is everything.
The Technical Challenge I’m Investigating
This isn’t just about building another chatbot. It’s about exploring whether Small Language Models can create genuinely personalized AI experiences that:
- Run locally
- Keep data private
- Actually learn from individual usage patterns
The questions I’m working through:
- Can SLMs be trained on personal data without cloud dependencies?
- How do you design voice recognition that works with older adults?
- What does photo recognition look like when it improves through family input?
- How do you make context retrieval feel natural, not robotic?
Solving these problems doesn’t just help my dad. It proves a model for how AI should work: personal, private, and genuinely helpful.
Why I’m Sharing This Journey
I could explore this quietly and reveal it when it’s polished.
But that’s not how innovation happens anymore.
People want to see the process, not just the result. They want to understand the decisions, the failures, the discoveries.
And frankly? I want to be held accountable.
When you announce you’re exploring something, you have to actually explore it.
What This Means for My Career (And Maybe Yours)
This exploration isn’t just about creating a prototype.
It’s about demonstrating a different kind of expertise.
- Instead of writing about AI’s potential, I’m testing what’s actually possible.
- Instead of analyzing other people’s products, I’m prototyping my own.
- Instead of just having opinions about technology, I’m experimenting with it.
This shift from observer to creator isn’t unique to me. It’s happening across industries.
The people who thrive in the AI era won’t just be the ones who understand the tools.
They’ll be the ones who can shape them.
The Real-World Test
Here’s what I’m working toward:
My dad should be able to look at a family photo, ask “Who is this?”
— and get back not just a name, but a story. Context. Connection.
If I can make that work, if I can prove that personalized AI can help someone maintain independence and dignity, then I’ve created something meaningful.
And if I can’t?
Then I’ve learned something valuable about the gap between AI hype and AI reality.
Beyond mem.ry
This exploration is bigger than one prototype. It’s about proving a philosophy:
- AI should be personal, not generic.
- AI should be private, not surveillance.
- AI should enhance human connection, not replace it.
Whether mem.ry becomes a product, attracts collaborators, or becomes a valuable learning experience, the process of building it is already changing how I think about:
- AI
- Product development
- My own career
What I’m Learning Already
We’re only in the planning phase, but I’m already seeing things differently:
- Writing about AI and building with AI require completely different skillsets
- The technical challenges are harder than I expected, but more solvable than I originally feared
- Having a personal connection to the problem changes everything
- Building something real forces you to make decisions that writing about concepts never does
The Invitation
If you’ve been learning about AI, analyzing AI, or teaching AI, but not building with it. Maybe it’s time to shift gears.
The conversation is moving from:
“What can AI do?”
to:
“What are you doing with AI?”
And that’s a much more interesting question.
- The technical decisions
- The user testing
- The failures and breakthroughs
It’s not that I think mem.ry will change the world. It’s that building it is already changing mine.
And in a landscape moving this fast,
that perspective shift might be the most valuable thing of all.
The future gets built by people willing to move from theory to practice.