I recently left my role at Superlogic.
I am not leaving technology. I am not stepping away frombuilding. And I am not taking a traditional sabbatical.
Instead, I am using this next chapter to focus on my family,help build a nonprofit, and teach my kids how to createreal products and businesses with AI.
That last part is the experiment I am most interested in.
My kids are growing up in a world where the cost of turningan idea into working software is dropping fast. A motivatedperson can now move from concept to prototype in days,sometimes hours.
But faster development does not automatically produce betterproducts.
AI can generate code, designs, copy, research, and businessplans. It cannot reliably decide which problems are worthsolving, whether users actually care, what tradeoffs areacceptable, or when a product is ready to ship.
Those are the skills I want to teach.
## The goal is not to teach them prompting
I am not trying to train my kids to become professionalprompt engineers.
I want them to learn how to build.
That means learning how to:
notice real problems
talk to potential users
test assumptions
define a small first version
use AI to accelerate execution
inspect and challenge generated work
launch something publicly
measure what happens
improve it or shut it down
The AI tools are important, but they are not the curriculum.
The curriculum is judgment.
A child who can ask an AI model to generate an app has donesomething interesting. A child who can explain why the appshould exist, who it serves, how it might make money, whereit could fail, and what should be built first has learnedsomething much more valuable.
## Senternet is our workshop
I have launched Senternet, aproduct and consulting studio. The name is not new:Senternet was my first company, founded 27 years ago, andreviving it now feels right.
It is the umbrella under which I will build products, createapps, advise companies, and experiment with new businessideas. My kids will participate where it makes sense, not aspassive observers, but as contributors.
That may involve:
researching a market
naming a product
interviewing users
designing an interface
creating marketing assets
testing a prototype
reviewing analytics
helping decide what to build next
Some projects will be client work. Others will be productswe own.
Some may become businesses. Some will fail quickly. That isexpected.
I want them to experience the entire loop from idea toexecution, including the uncomfortable parts: unclearrequirements, bad assumptions, bugs, rejection, and userswho behave differently than expected.
AI makes it easier to build the wrong thing faster.
The only defense is learning how to think.
## We will use AI, but we will not outsource responsibility
My background spans engineering, security, operations,product, design, and executive leadership. I have worked asa CEO, CTO, COO, CISO, founder, and builder.
The most useful thing I can teach my kids is not a specificframework or programming language. Those will change.
I can teach them how to reason through systems.
When we use AI to generate software, they will need to ask:
Does this code actually work?
Is it secure?
What assumptions did the model make?
What data are we collecting?
Who owns that data?
What happens when the API fails?
How much will this cost at scale?
Are we solving the original problem or just adding
features?
Would someone pay for this?
Should this exist at all?
The model can produce an answer.
The builder still owns the consequences.
That distinction matters, especially for kids who mayotherwise grow up believing that plausible output is thesame thing as correct output.
## Bee Ready gives us a real-world problem to solve
I will also be volunteering with BeeReady, a nonprofit co-founded bymy wife, Andi Senter, another physician mom, Elda Fisher,and me.
Bee Ready is focused on improving emergency preparedness atyouth sporting events through CPR and AED training, visiblevolunteer response teams, and better access to lifesavingequipment.
I will serve as CTO and COO.
That means helping with technology, operations, internalsystems, volunteer coordination, data collection, and theinfrastructure required to grow the organization.
This gives us something better than a classroom exercise.
It gives us real constraints.
Nonprofits have limited budgets. Volunteers have limitedtime. Users may be stressed, distracted, or nontechnical.Systems need to work at fields, pools, and community events.Software cannot exist just because it is interesting tobuild.
It has to reduce friction.
It has to support the mission.
And in this case, failure can matter.
That makes Bee Ready an unusually meaningful environment forteaching product development, operations, and responsibletechnology.
## The stack matters less than the process
We will use modern AI-assisted development tools. That willlikely include coding agents, design tools, automationplatforms, cloud services, analytics, and traditionaldevelopment environments.
But I do not want this project to become a running list oftools.
Tools change too quickly.
The repeatable process is more important:
Find a real problem.
Define the user.
Identify the riskiest assumption.
Build the smallest test.
Use AI to accelerate the work.
Review everything critically.
Put it in front of real people.
Measure behavior instead of collecting compliments.
Decide whether to continue.
That process works whether the product is a mobile app, anonprofit workflow, a consulting service, or a small onlinebusiness.
It also prevents AI-assisted development from turning intoendless prototype generation.
Shipping is not the end of the process.
Shipping is when reality starts grading the work.
## I expect them to fail
I do not expect every project to succeed.
I would be concerned if they did.
A project nobody uses can teach positioning.
A product nobody buys can teach pricing.
A confusing onboarding flow can teach design.
A security mistake can teach threat modeling.
A feature that takes three weeks and adds no value can teachscope control.
The point is not to manufacture a string of impressivelaunches for social media.
The point is to help them develop the ability to recover,adapt, and keep building without confusing failure withpersonal inadequacy.
AI reduces the cost of experimentation. That should make usmore willing to test ideas, not more attached to every ideawe generate.
## This is also an experiment in education
Traditional education often separates disciplines.
Programming is one subject. Business is another. Writing,design, finance, operations, and marketing live somewhereelse.
Building a product combines all of them.
A small software project can require a student to writeclearly, think mathematically, understand users, evaluatetradeoffs, manage time, communicate decisions, and acceptcriticism.
AI can help at every stage, but it also creates a neweducational problem: students can produce work they do notunderstand.
So one rule will be simple:
You should be able to explain what you built.
That does not mean memorizing every line of generated code.Professional developers already rely on frameworks,libraries, abstractions, and tools they did not write.
It means understanding the system well enough to reasonabout its behavior, limitations, risks, and purpose.
If you cannot explain why it works, you are not done.
## What I plan to write about
I intend to document this experiment.
That may include:
how we choose projects
how I teach kids to use coding agents responsibly
where AI-assisted development works well
where it creates hidden problems
how we validate product ideas
how we structure small family projects
lessons from building nonprofit technology
mistakes we make
products we launch
projects we decide to kill
I am especially interested in the gap between generatingsoftware and building a business.
That gap is still enormous.
AI is making implementation cheaper. It is not makingcustomer understanding, distribution, judgment, leadership,or accountability obsolete.
In many cases, it is making them more important.
## A different kind of career move
Leaving a day job is often framed as a dramatic leap intoentrepreneurship.
This feels different.
I am not betting everything on one startup. I am building astudio, helping operate a nonprofit, spending more time withmy kids, and creating a place where we can test ideastogether.
For most of my career, I have helped companies buildproducts and solve technical problems.
Now I want to apply that experience more directly, whileteaching the next generation of my family how to createrather than simply consume.
We will use AI heavily.
We will also question it, test it, reject its output, andtake responsibility for what we ship.
That is the part of AI-assisted building I think mattersmost.
Leave a Reply