Skip to content

Week 4 at Antler: I’m Building an AI-Powered Browser for Kids

Posted on:May 20, 2025

Week 4 at Antler: I’m Building an AI-Powered Browser for Kids

“The way to do zero to one product development is to execute at 100% for the thing you’re trying to validate at that specific stage of the product development cycle. And then the rest can be half assed” - Nikita Bier

The idea here is to avoid scope creep as aggressively as possible because you have neither the time nor the resources to do many of the “important” things you want to build.

At the same time on the other side of that are the assumptions we are testing against.

Constantly truth-finding.

And if that assumption is

false

…

What did we miss? What did we overlook? Can we avoid overlooking that next time?

If this assumption is

true

…

What other assumptions need to be tested?

For us we are working on a refined version of my real time voice/screen capturing idea, we are helping parents capture their kids screen time and inform them where their kids spend most of their times, how they learn and alert them in real time in the case of

potential risks, inappropriate content, or signs of distress

.

If your child shows signs of chronic distraction or is a victim of cyberbullying, you want to know—not after it happens, but as it’s happening.

Despite Screen Time APIs and VPN-level parental controls, there’s still a black box around

what kids are actually consuming

. Parents can see screen time, but not

meaningful insight

—is it Bluey or YouTube drama? Is it syllabus content or TikTok trends?

What cartoon are they watching? How many times they loose focus and switch back and forth different channels, how much of this content is their syallabus, how much are topics they should be knowing about etc.

We have a couple of ideas;

My earliest memories using computer was with Encarta, it was a portal to into this whole new world. I know there is a teenage angle

The VPNs will never just be able to get all that data we need.

[The 1,000th Application

Femi Awoniyi

2 months ago](https://www.linkedin.com/pulse/1000th-application-femi-awoniyi-eydfc)

[The Art of the Pitch (Start with a Bang, End with a…

Kayode Oyewole

9 years ago](https://www.linkedin.com/pulse/art-pitch-start-bang-end-kayode-oyewole)

[Tech recession or reality check?

Temitayo Diadem Adegoke. DBA.

3 years ago](https://www.linkedin.com/pulse/tech-recession-reality-check-temitayo-diadem-adegoke-)

Browser extensions are able to do more, simply because its in the browesr and can track and measure user behaviour too a great degree. Relatively easy to build versus a browser.

Finally

the browser. 90% of the type context we want to capture can be rendered in the browser. This means that we can clone a browser, keep it simple and remove features that we don’t think is necessary in a child browser, add all the things that make it AI powered and kid safe and start working towards PMF.

This product isn’t a hard sell—because the pain is real. However this particular market is tricky. I and my teamate at

Antler

discussed this briefly

A serious issue with our market is primarily attention.

Parents are overwhelmed and distracted. For us to have the network effects to take over apple store for example or go viral on tiktok or something similar we’d need to build a product that is so good it forces users to tell other users. In our case it’d be parents telling other parents. Teenagers naturally share apps with each other, parents not as much. So it will be hard to pull off virality. A healthy LTV:CAC ratio is generally considered to be around 3:1 or higher.

Virality means we can subsidize our CAC and give us some gunpowder to invest resources into building a more superior product for the market.

The LTV of this market is between

24–30 months,

to get this number I am using mainstream parental control apps as a proxy. If we do things right we can eventually offboard users into a browser. And what does a browser give us ? More user context and eye ball space to sell… Ads.

Brands (like

the LEGO Group

, Crayola, Nat Geo Kids ) can sponsor

mini-games or interactive learning pages

. Duolingo can pay to promote its new kids feature in the “learning dashboard”.

On the parent dashboard outside the kids browser we can deliver key insights and recommedations like “Your child spent 2 hours on math content—here are math kits parents are buying.”

This week we’re validating the assumption:

‘A browser extension can give us enough context to meaningfully classify the content a child is consuming online.’

If true, it unlocks an early PMF wedge. If false, it pushes us closer to needing a full AI-native browser.

The good people at

The Browser Company

Josh Miller (CEO) and Hursh Agrawal (CTO) started building a browser in 2019, on 2024, December they Publicly announced as “Dia,” with a teaser video and new website, of course they are going for a mainstream audience, from what I have seen it looks like they want to make AI native to the browser. I am positive that is the right approach, this will kill so many startups in a fell swoop. A metric I can get behind if you’re building something as mainstream as this.

How-many-chrome-extensions-micro-startups-indie-businesses-are-you-killing-at-once?

Here is the thread of the founder from a couple of days ago

https://x.com/joshm/status/1922639507033149597

they tested with students, some were using it to create flashcards from their open tabs and study material and I can not count on one hand how many companies that will oblierate once that launch.

If you’re a parent and would love to test out what we are building this week please DM me !


Originally published in Blackbox Thinking on LinkedIn.