Issue #25: A (Different) View

Edtech through a pair of product sunglasses.

JORDAN BOHANNON

Hello, it’s been a while. The View used to be about bootcamps. Going forward, I’m going to make this a shorter email: 5 things in edtech, through a product lens. I’m not going to commit to sending this every week, who has the time to worry about societal collapse and send a weekly email?

But I am committing to sending this out when I have 5 things in a Google Doc.

Let’s get into it.

Five Links

Read: How to use Google’s new Ideate Agent for Product Discovery and Ideation

Stitch has been an under-the-radar tool for PMs. I can see this update getting more visibility. I spent a couple of days playing around with it. It’s genuinely fast. What used to take 1-2 days in a design sprint now takes an hour. It feels like validating good ideas is now the bottleneck, not generating ideas. You just have to be careful about making assumptions around personas. Stitch might convince you they are real people.

Read: Spite House: AI, disintermediation and the end of the free web

A long read from Lauren Pope about how AI companies are upending SEO and content distribution. If your product relies on content, this is a genuinely terrifying read. If your distribution model depends on SEO (curriculum platforms, learning libraries, content marketplaces) you’re probably already seeing your traffic decline. Pope explains why the companies that survive will be the ones who own the learner relationship.

Read: Tackling the graduate employability crisis

Matt Walton (ex-FutureLean CPO) has an open-call out for collaborators on graduate employability. This topic is all over the news right now, with startling data around entry-level unemployment. However, IBM just announced they are doubling down on graduate-hiring. I guess nobody really knows what’s going on. Most products I’ve seen in this space optimise for placement over match quality. Interested to see how this develops. I worked with Matt in a past life and his archive of Product Learning case studies are a must-read for anyone building in the space.

Watch: “Multiverse is growing faster than ever”: Euan Blair, Founder and CEO at Multiverse - YouTube

A surprisingly candid interview with Multiverse CEO, Euan Blair. Touching on the acquisition of German startup, StackFuel. I was there (“I’m losing my edge”) in 2020 when we chose the US over Europe. It felt like the bigger market. Turns out selling apprenticeships without government subsidy is very hard. This German pivot is an shows that subsidies can matter more than TAM. Not many views on this video which is surprising as it’s probably the most insightful MV interview I’ve seen in a while.

Read: Breaking the Fourth Wall of Fourth Spaces

Demand for live events is up 292% since 2021. People are looking for friendships that don’t happen on a screen. Gen Z and millennials drink less but still want to gather. For example, Felecia and Tyrone Freely built ‘Lectures on Tap’, an event company hosting lectures in bars. People are burned out on Zoom education and in-person is ripe for a comeback. If you’re building a purely digital learning product, think about what your IRL looks like.

Going Deeper:
Listen: Inside Alpha School: An AI-Powered School

Alpha School, an “AI-powered school” is generating lessons that do “more harm than good”, according to a 404 Media Investigation. The Trump administration has lauded the school, presumably because their vision is to get rid of teachers “run more efficiently”.

This isn’t a hit piece. The lesson for PMs is that Alpha School is trying to pivot to being an AI company. And running into some common issues.

1. Hallucinations. The AI-generated lesson plans generate faulty questions and answers 10% of the time. The school uses AI to vet the quality of AI-generated questions. You don’t need to be a genius to see this leads to degraded quality over multiple loops.¹

2. Content scraping. There isn’t enough original content to differentiate the school’s curriculum Alpha has scraped content from other education platforms to incorporate into their own lessons, without permission (employees signing up to external platforms, scraping from textbooks without attribution).

3. Surveillance technology. A story as old as time with edtech, but there’s a large amount of surveillance tech on children going on (mouse tracking, website visits, how much time spent per task). It’s positioned under the guise of improving the learner experience but Co-Founder Jo Liemandt also runs spyware company CrossOver. Is it a product need to surveillance-capitalism children?

Don’t read this as a K-12 story. These are the same problems surfacing in corporate learning and edtech too. The school’s product vision is “no-human-in-the-loop”. Seriously.

The uncomfortable caveat in the reporting is nobody disputes their claim that students score in the top 2% nationally (although presumably as a parent that’s what you’d expect for $65k tuition). The gap between AI hype and AI execution is real, but the story isn’t as simple as “AI bad”. Unfortunately, it is children who are “being treated like guinea pigs”.

One more thing:
Been feeling down all week. Chalk it up to February, remote-work, and a cold that won’t shift. Also, maybe every article being about AI anxiety doesn’t help.

This is from Derek Sivers’ 2021 pulls you up short: Here’s how to live: Master something. A little corny? Perhaps. But sometimes corny works.

Mastery is the best goal because the rich can’t buy it, the impatient can’t rush it, the privileged can’t inherit it, and nobody can steal it.
You can only earn it through hard work.
Mastery is the ultimate status.

Striving makes you happy.
Pursuit is the opposite of depression.
People at the end of their life, who said they were the happiest with their life, were the ones who had spent the most time in the flow of fascinating work.

Concentrating all of your life’s force on one thing gives you incredible power.
Sunlight won’t catch a stick on fire.
But if you use a magnifying glass to focus the sunlight on one spot, it will.
Mastery needs your full focused attention.

Back soon.

1. Although the Feb 5th release of GPT -5.3 Codex from OpenAI claims Codex is “the first model that was instrumental in creating itself. The Codex team used early versions to debug its own training, manage its own deployment, and diagnose test results and evaluations.” Newer models hallucinate at a lower rate. Although do we know that’s zero?