Idea to Startup: A Live Idea Test, Part Two - AI for Parenting

Podcast | Youtube

Sponsor: Tacklebox

20% off monthly, 6-mo and year-long subscriptions to the Self-Paced Membership. If you’re struggling finding the path to follow to start your idea, this is the way.

code: MuchAdoAboutStuffing

This Episode

Today is episode two of testing an idea (AI for Parenting) live on the pod. We use a second round of interviews to go deep on the actual problem we're solving for parents, pull inspiration from an AI tool in the dementia care space, and end up with a Wedge product that'll use voice notes to reduce the pain of handoffs. We also hit on one of my favorite tactics - The Pain Text. Hot Frosty is in there, too, because why not. 

Pod References

• Tacklebox Customer Interview Workshop

• You And Your Research - Richard Hamming 

• How to Find Your Wedge

00:30 Intro
01:20 Part 1 Recap
06:00 Picking the Problem
07:44 Interviews Round Two
10:54 The Text Prompt Tactic
14:51 What are the Stakes?
15:52 Customer Interview Workshops
16:20 The Wedge
21:01 Problems and Pain
22:20 The Stakes, Part Two
23:30 What’s Next

Transcript - Feel Free to Read it Like a Long-form Article:

Part two!

Today, we’re jumping into part two of us starting a startup live on the pod. We uncover a secret about parents maybe good enough to support a business, we build an AI wedge product to test it, and we go into detail on all the hard things people tend to screw up when going from idea to the first thing you put in your customer’s hands.

In the words of the strangest character from that old frosty the snowman Christmas cartoon that, if we’re lucky 10% of you have watched - we’ve been busy busy busy

< clip>

The hope, as always, is that hearing me do all this will make it more approachable.

Now, if you missed part 1, you probably don’t have to go back and listen to it to if you don’t want to. I’d love it if you did, but I’ll give you a firm base now if you were too busy watching hot frosty and drinking egg nog the past two weeks to tune in. And if you haven’t watched hot frosty, don’t. It’s not bad enough to be good or good enough to be good, if that makes sense.

Anyway, on to the recap.

Previously, on idea to startup.

Last episode, we kicked off a series starting an idea live on the pod. The idea was AI for Parenting, and we chose it because lots of people are applying to Tacklebox with a similar type of idea - AI for “insert your customer here.”

Over the years we’ve gone through waves of blank for blank ideas - we had the airbnb and uber for’s… airbnb for lawn equipment or uber for chefs, then it was warby parker for office furniture or toothbrushes, then subscription boxes for cooking spices or men’s face cream, then people tried to gamify everything - from laundry to kids chores to grocery shopping to maybe the boldest Tacklebox app of all time, a startup that gamified sex called…. game of moans. Sheesh. The literal idea was to have a feed like fitbit where instead of tracking steps against your friends you’d have a couples account where you’d track - I mean, you get it.

Anyway, I generally hate these types of ideas because people misunderstand the hard part. They think it’s thinking up the thing that comes before the “for x.” When pitching their idea they’d dramatically say something like, “get ready for this… airbnb for… lawn equipment,” then confidently lean back in their chair like they were billy mays.

The hard part of Airbnb for lawn equipment isn’t thinking it up - it’s building the marketplace. Creating trust with first customers, finding the early use cases where each side feels disproportionate pain at the exact same time, facilitating an interaction where one side happily overpays for a lawnmower because they can get it right now, figuring out how to amplify that transaction so that it creates three more without you needing to spend a dime on marketing. The lawnmower and airbnb part of it are basically irrelevant.

You can build a successful business in one of two ways:

You can be the best entrepreneur in the world and pick just about any idea and simply out-execute everyone, or, you can be a normal person who picks an idea where you’ve got asymmetric information.

The magic trick you pull is either you actually needing to read someones mind to guess which card they drew from the deck, or, just strategically placing a mirror behind them. I choose the mirror. Asymmetric information.

Which brings us back to last week’s episode and why I’m, maybe surprisingly, excited that everyone is pitching me big, broad ideas like AI for carpenters even though we’ve established startups require asymmetric information and that sentence shows none.

The reason is that the AI part of the pitch is helpful in a way the airbnb or warby parker part of the pitch never were.

AI can remove barriers for testing and getting feedback and can speed up the critical “find asymmetric information” stage. It can replace stuff that used to be time and resource intensive, like getting something potentially useful in front of customers.

The magic will still come from the interactions - the uncomfortable, un-outsourceable conversations and ethnographic research with potential customers. But getting those interactions and building Concierge MVPs that can cull the type of feedback that leads to asymmetric information is possible today in a way it simply wasn’t 18 months ago.

The phrase “AI for carpenters” isn’t the end, it’s the start, but new tools can help you go from “AI for carpenters” as a broad idea to “a tool that evaluates project parameters and provides accurate cost and time estimates for one-person businesses” fast. And a non-carpenter, non-technical person can find their way there. I know because that’s a real example from a company in Tacklebox.

Pursuing and testing an idea is way more accessible than it used to be.

So, feel free to use AI for X as a permission structure to get started. We’ll teach you how to niche it down. Which is what we do in this series and with our example - AI for parenting.

During the first episode, we started broad - literally, with the phrase AI for parenting. Parenting is hard, it’s important, and it’s changing, as there are far more households where 2 parents work than there were 20 years ago.

I ran a bunch of interviews with friends and friends of friends who are parents to problem hunt - to find the specific hurdles these parents encountered that they’d spent time and money and sleepless nights trying to navigate.

From these initial conversations, four juicy problems jumped out. Problems that were urgent, expensive, painful, growing and frequent.

1. Logistics - how to get kids to the right places at the right times,

2. Development Uncertainty - or, is it normal for my kid to be doing X or not doing X yet,

3. Sleep training - basically, get my damn kid to sleep, and

4. Care Coordination, - how to keep everyone on the same page when a bunch of people are handing off taking care of your kid throughout the week.

We landed on the last one - care coordination. In households where both parents worked and had a kid or kids under 5, this was a core problem. There was often a mishmash of care- maybe a grandparent taking the kid monday and tuesday, a different grandparent on wednesday, a nanny on thursday and the parents cobbling together friday, or some other patchwork solution. Keeping everyone in the loop on food and sleep and development and appointments and everything else was hard and, as we’ll see, stressful.

There were a number of reasons to start with care coordination, But, as good as any, was.. speed. It felt actionable and it seemed like interviewees were already actively trying to solve it and entrepreneurs have to make lots of decisions with incomplete information and speed to that decision is critical. So we’ll sprint at this for a bit then pick up our heads and see where we’re at.

Once we decided on a problem, we ran a second round of interviews zooming in on Care Coordination, which we’ll recap now. After that, we’ll do a little smooth jazz and then finish with the concierge MVP we built in a weekend based off those deeper interviews.

Let’s get into em.

—

Interviews Round 2: Earned Wisdom + Secrets

My old boss was enamored with what he called “Earned Wisdom.”

When founders would come in and pitch a product that, say, partially relied on content marketing for a rollout strategy, and they didn’t have any content marketing experience, he’d give them homework. To earn a bit of wisdom.

Go get a few lumps, earn a perspective on content marketing, then come back to me, he’d say. He’d then jump on their side of the table and help them figure out how to earn that wisdom. Maybe it was a set of campaigns they could run or a freelancer they could hire. The goal was to get your hands dirty through a quick, targeted project.

He used to tell founders that the thing they were avoiding - the thing that was intimidating or uncomfortable or unknown - would become manageable with a little earned wisdom. He LOVED founders that’d done unusual things to earn wisdom - one founder had gotten a job at an urgent care clinic for a year to understand the inner workings before starting her business. My boss invested almost immediately.

Round two of interviews for the AI for parents idea was all about earning a bit of wisdom.

The first round of interviews had been lighter - we’d talked about lots of problems around parenting - but the second round would be far more targeted. I wanted to get hyper specific around how these parents coordinated care, and, more importantly, how they felt at each step.

I emailed 10 people who’d mentioned care coordination as a problem asking if they’d be willing to go deeper and onlye got 4 responses. I considered sending another round of cold emails to try get more interviews, but decided on running these four first to see if there was anything sticky I could turn into a test. Parents are busy - a good data point for thinking about customer acquisition.

To kick off each interview, I set a distinctly different tone from the first chat - “I’m no longer looking to avoid anchoring you,” I began, “my goal today is to go really deep on one problem - to hear about how you manage it, what you’ve tried, who you’ve spoken with about it, and how you feel during each stage of it.”

Almost immediately, each person I spoke with went deep on, what an interviewee called, information lag.

“Let’s say the kid was working on putting his shoes and socks on at daycare,” they said. “Our grandparent who comes Monday might know this, and work with him on it. But whoever is covering Tuesday might not. So, sometimes people make him take off his shoes and sometimes people take his shoes off for him and now he’s confused and when he’s confused he gets frustrated and cries.

This might sound insignificant, but he’s learning like 30 things a day. Kids develop fast. So what’s expected of him changes constantly and he’s always frustrated. If you come once a week, you can’t keep up.”

I heard this sort of thing from all four.

I asked each if I could do a little ethnographic research - if I could watch the kid handoff. Not surprisingly, everyone said no and seemed semi-creeped out, which, when I stand back, I obviously get. But, two agreed to call me or text me the next time they felt stressed about the care situation.

This text prompt is one of my favorite strategies with people you interview who seem like great potential customers. It starts to move them towards being a design partner. Eventually, we want our product to feel magical - meaning, when our customer reaches the hard part of their process, we jump in and handle it. For now, them calling or texting us during the hard part will give us deep insight into how we might build for the moment. This strategy has anchored a bunch of the best businesses that have come through Tacklebox.

One interviewee suggested something I should’ve thought of - “hey,” they said, “do you want to interview my aunt? She takes my daughter on Wednesdays and I’m sure has some thoughts.”

I did. And the story from the Aunt’s perspective was fascinating. She’d get to the house and try to avoid asking too many questions because she didn’t want to seem incompetent. But, she said, this meant she was never totally up to date. She was always guessing - did he still nap at the same time? Was he using a fork yet?

After the four parent calls and the one aunt call, I tried to synthesize what I’d learned.

My first thought, and probably yours, too, was the obvious communication problem. There isn’t a running log or diary of what, exactly, the kid is up to at that moment.

But… this wouldn’t be hard to solve. Every family I spoke with had some kind of notebook or text chain or something that gave the day’s directions. The parents could easily write what the kid was focused on or the caretaker could ask. It just didn’t work.

So… why?

It became clear through two separate quotes. One from the aunt, and one from a parent.

Here’s the aunt first:

“So, I take care of the kid one day a week and my mom - his grandma - takes care of him once a week, too. And we always talk about this - what are we supposed to be? Am I the fun aunt? I’d love to be the fun aunt, but I’m also the main parent 15% of the week. My mom says the same thing. Is she a grandma? Or is she supposed to be a mom?

And, neither of us like being treated like an employee. We’re family.

I think the big problem we struggle with is our role. What… are we? Am I an aunt? A mom? An employee? What?”

And now, the quote from a parent I spoke with about getting help from a grandparent.

“I feel like there’s this push back from all the grandparents involved. It feels like we aren’t on the same team - they’re telling me how they raised kids and they don’t need help or advice or input, and I’m trying to tell them how I want my kid raised while still acknowledging that I appreciate their help and know they have a ton of experience. And, while I’d love to write out a bunch of rules, I just can’t do it. I hate doing it. I don’t want to tell my parents how to parent. It’s just a mess. It’s really hard. But we can’t afford a full-time nanny and neither of us can quit our jobs.”

Woooo boy! There are some emotions here!

The real problem, I think, isn’t just about communication—it’s about an identity crisis. Family caregivers feel stuck between roles: they can’t fully embrace being the fun aunt or favorite grandparent while enforcing rules, but they also don’t want to be treated like employees because… they’re family. This role tension is creating stress that no calendar or group chat can fix - at least one that isn’t built with this tension in mind.

This is a hard problem. It might not be solvable. But hard problems you aren’t sure you can solve are the only things worth going after.

And, that leads to a simple questions about the problems you choose to solve.

What are the stakes?

What happens if this problem doesn’t get solved?

There’s a quote by Richard Hamming that I love - and if you haven’t listened to or read his talk You and Your Research, I’ll pop it in the show notes. It’s outstanding.

The quote goes:

“If you do not work on important problems, how can you expect to do important work?”

In each of my interviews I came away thinking about how important this problem was to solve. For the kids development, but also for the parent’s mental health. It’s hard enough balancing raising a kid with a demanding job, but adding in the dynamics with family members amplifies the whole thing. A number of people I’d interviewed had already talked about or gone to therapy to try to deal with it all.

The stakes are high. This thing is worth solving or at least chipping away at.

So, let’s get some more earned knowledge. Let’s try to run a test.

What’s that look like, though? What’s a wedge product? What do we… do?

Tricky tricky.

We’ll take a stab…after… a little smooth jazz.

The Wedge

We’ve talked about wedge products before and they’re probably what listeners reach out to me about the most.

If you aren’t familiar, a wedge product tackles a small, manageable piece of a bigger problem for your first customer. It’s got a short feedback loop with clear success metrics. The goal is to earn some wisdom - to help a customer through the first step of a deeper problem. Wedge products often end up as your main top of funnel acquisition tactic.

If you want more on wedges, I’ll link to a few pods we’ve done on them in the show notes.

A key component of a good wedge product is clarity around exactly what success looks like. How it’s measured and how it feels.

So, starting with your hypothesis of success is a good place to start.

My current hypothesis is that success for the patchwork childcare coverage problem will end up with everyone feeling like they’re on the same side of the table helping out with the kid. Information is consistent and delivered in a way that makes everyone feel valuable, with clarity around roles and expectations.

That same old boss I talked about before used to say that most people products succeed by doing one of two things really well: getting people to the same side of the table or making them compete directly against each other. Trying to do both—or being unclear about which you are— sinks the ship.

To get everyone to the same side of the table we’ve got to handle the tension.

Wedges work well when they focus on a specific, highly charged moment.

The moment I’m choosing is the handoff - when a caretaker takes over for a parent, and when the parent returns.

The reason I chose this moment is it’s when parents texted me saying they felt the pain. They’d hand off their kid, get in the car to go to work, and text. “I didn’t tell them about x y and z because I didn’t want to overwhelm or overstep, but now I wish I had,” or “I was mean and snippy for no reason” - snippy was a direct quote and I’ve adopted it. Might be the word of 2025 for me.

Anyway, I needed to zoom in on that moment and help with communication and clarity of roles.

AI could, theoretically, play a big role here.

As we mentioned in the first episode, AI needs to be a response to the specific problem your customer has. So, during the lulls, as I waited for my interview calls, I explored similar-ish use cases that had been solved recently with AI. A new-ish company that’d gained traction in the dementia care world seemed like a good fit.

The product dealt with a similar dynamic. Dementia patients who had a number of caretakers that were reporting back to loved ones and doctors.

At the end of each shift, the product prompted a caretaker for a person with dementia with a number of questions - there were questions about how the day went, then specific questions about the exact amount of food eaten and when, sleep times, and behaviors, and finally open ended prompts where the caretaker was invited to just talk through the full day. The response was captured like a voice note. The app then summarized the report for a few stakeholders. First, it’d create a pertinent update for the parents. Next, it’d update the food, sleep and behavior log. Finally, it’d create a document doctors could view to see the effect of, say, a change in medication or schedule.

I figured we could do something similar.

I could text parents and caretakers a set of questions they could answer longform in a voice note. I could then transcribe and summarize this via an AI tool like Claude, then create progress reports or goals or whatever else the parents needed.

This would reduce the friction for gathering critical information.

The other part of the problem is sharing that information. The AI tool for dementia patients changed the tenor and information based on their audience. It talks to doctors like doctors and loved ones like loved ones.

So, we could try having the messages to grandparents feel different from the messages to parents, maybe.

This might sound a bit out there, and maybe it is. But, communication style is critical. In nearly all of my interviews, communication was a huge blocker. Parents told me the worst part of their day was trying to tell caretakers what they needed them to do. Removing the communication onus from the parent could be pretty helpful.

And this brings us back to problems and pain.

People only solve problems after pain, and people only actively solve one of two types of problems. What I call hole problems or teleporter problems.

Hole problems are what you’d think. Someone fell into a hole and they need help getting out. The world doesn’t spin until they get out. It’s priority 1 2 and 3. The care communication problem is not a hole problem.

The second type of problem worth solving is a teleporter problem. There’s a process your customer does every day. The process has a bunch of steps. You find the hardest steps and remove them. They teleport from step 3 to step 7, and you handle steps 4-6. The more painful that process, the more frequent that process, the more they’ll pay for you to take it off their plate.

So… my hypothesis is that the hardest steps we can remove are the communication steps.

So, I sent an email to a few of the parents I’d spoken with asking if they’d like to partner up to try and solve this problem. I said that when they reached that tricky point - the handoff - we’d take over.

Some people were excited, some people didn’t respond, but we got two design partners willing to give it a try.

And now, a quick note on, again, the stakes.

This is a big, hairy, important problem. And I’m not totally sure I’ll be able to solve it with whatever I pull together for this wedge product. And, I definitely feel hesitant to get involved in these potentially messy situations.

So, how do I move forward not being confident I can create a great solution for these customers?

Scope, depth, and clarity.

I will, at most, be working with two families to start. Each will be what’s called a design partner - I’ll make it clear I have deep understanding about the problem they’ve got and what success would look like, but also that the tools I’m using are new.

And, no matter what the initial, quote, product looks like - this will be extremely manual. I’ll be keyed in on every step of the process so that no communication would be sent to, say, a grandma without the parent first seeing it.

Your first product should feel like this. A bit scary, but, not out of your control.

In the next episode of the series, we’ll talk through the details of the wedge product test. I’ll go into the big questions - did we charge for it? How’d we build it? What happened?

And, we’ll probably give the evolved punny name.

These episodes are always tough to wrap up, because it’s really just a holdover to the next episode.

But, if you’re in the early stages of your own idea, here’s what you should probably take away:

1. Start with customer interviews to find a problem with serious stakes. Ones that might be worth your time.

2. Zoom in on that problem and devote a few weeks or a month to it to start, even if you aren’t 100% sure it’s the right problem. Making decisions without enough information is the way of the entrepreneur

3. Find the specific moment you want to live in - the one your customer texts you about because they feel the pain - that’s the focus of your wedge product.

4. Search for products out there solving similar-ish problems in different industries well for inspiration.

5. Feel like this whole thing is a bit out of reach. Like, it should be solved, but maybe you aren’t qualified or ready or capable of solving it. That’s the sweet spot.

And then, relax for a night and watch Hot Frosty. I know I said it was bad but I keep thinking of it? Maybe it’s just the title? Maybe it’s bad enough to be good? Or just good?

I don’t know. Maybe we’ll figure it out along with the wedge product in the next episode.

And, also, if you got this far and feel like you’ve got this problem - reach out. I’ll take some more beta testers. Team at gettacklebox dot com.

Have a great week.

This was the idea to startup podcast brought to you by tacklebox. Start your startup in 2025, and do it with people who can show you the way. gettacklebox dot com.