Stop Treating AI Like a Search Engine: How to Build With It Instead (with Sarah Heeter)
listen wherever you get your podcasts:
The Market Viability of One
Why you should build the tool instead of buying it
You know the feeling. You sign up for a piece of software because it does the one thing you need. Then you realize you are paying every month for twenty features you never touch, a client portal nobody asked for, or a scheduling tool that cannot match your brand. You tell yourself it is fine because at least it works. Except it does not work, not the way you need it to.
On this week's episode of AI Queens, Sarah Heeter of Podfox Media joins Erika Stanley for a conversation that starts with a simple idea and ends up rewriting how you think about every subscription in your business.
The idea is this. Most people are using AI to answer questions. Fewer people are using it to solve problems.
“I try really hard to encourage people to use it to solve problems instead of just to answer questions,” Erika says. “That is the biggest compliment you can give me, that you are using it to do big things.”
That distinction matters more than it sounds. Asking a question gets you an answer. Solving a problem gets you a system. One saves you a few minutes. The other can save you a monthly invoice, a support ticket, or a season of missed appointments.
The commercial that gets it backwards
There is a Chat GPT commercial making the rounds where a woman asks the voice assistant if it is still raining outside and what time she needs to leave to catch her train. It is a perfectly nice ad. It is also a perfect example of underutilizing the tool. You already have a weather app. You already have a clock. Pulling up a chat window to ask a question you could answer in two taps is not innovation, it is a party trick.
The real shift happens when you stop asking AI to answer what you already know and start asking it to build what does not exist yet.
You have to retrain your brain first
Building with AI works a lot like building automations in Zapier. The first time someone hears about automation, they get excited, then they freeze, because they do not know what to automate. It is not until you build a few workflows that your brain starts seeing the world differently. Eventually you look at any repetitive task and picture the trigger, the action, and the outcome before you have opened a single tool.
The same thing happens with AI. The more you build, the more you start seeing your business as a list of specific, solvable problems instead of a list of tools you have not bought yet.
The skateboard, not the car without an engine
A minimum viable product is not a stripped down version of your big idea. It is the simplest thing that gets you from point A to point B. If you want to get somewhere on wheels, you do not start by building a car. You start with a skateboard. Then you add handles and it becomes a scooter. Then you add pedals. You build your way up.
It is easy to think you have already scaled back to your MVP when really you have just built a car with no engine. The real test is whether the simplest version still moves.
A real case study: the $15 a month problem
Sarah ran into this directly. She hosts small group roundtable sessions for her own podcast, capped at five people per session, built around a specific topic each time. She needed people to see her availability, pick a session, see the topic and time, add it to their calendar, and get a reminder.
The tool she was using charged fifteen dollars a month, would not let her control the branding, and could not reliably remind people about sessions they had already signed up for. People missed sessions they wanted to attend simply because the reminder never landed.
So she built her own. Using Claude Code, she described the problem and let the tool walk her through every step, including setting up accounts on platforms she had never touched before. She hosted it herself, connected it directly to her own website domain, and added no new monthly subscriptions beyond a few dollars for hosting. Before building anything, she also had AI research the market to confirm nothing already on the shelf solved her specific friction points. Nothing did.
This is not a story about becoming a developer. It is a story about being specific enough about a problem that a tool without judgment or ego can help you build the exact fix, one step at a time.
The personalized internet is already here
The bigger point underneath all of this is access. The general public now has access to the same underlying models that Fortune 500 companies use to build products. You do not need a boardroom of people deciding your problem is worth solving. You are your own market. If a tool solves your specific problem, that is the only viability test that matters.
Where to start
You do not need a big build to test this mindset. Start with the smallest possible version of a problem you already know well. Notice where you are paying for friction instead of a solution. Ask AI to research what already exists before you build anything. Then build the skateboard first.
This episode goes deeper into the specific tools, the trial and error, and the real numbers behind what these builds cost in time and money. Listen to the full conversation with Sarah Heeter on AI Queens.
