What is a bespoke tech startup?

I’ll be honest. I just made up that term.

I was trying to think of a way to describe slow craftmanship with the speed of technology. Here’s why: I don’t think everything should be mass produced. I don’t think technology should be used to replace quality work. And there are some things that should prioritize a deliberate, informed, long conversation such as where one is going to college. I think technology should be used to facilitate these conversations. They shouldn’t replace them.

Consider what happened when we started mass producing food. We became sicker and sped up the decline of the planet. We’re imploring people to eat food that isn’t so distant from what it looked like when it was growing.

Consider what happened when we introduced ed tech to replace teachers in a classroom. We’re wondering why kids are so anxious and impatient, lack conversation skills, can’t work in teams, and can’t think critically.

Consider what happened when we replaced everyday interactions with apps. We’re asking people to decide who they will marry based on a few sentences and curated pictures and our supposed town square is limited to 280 characters.

People have asked me what happens when thousands of kids are using the app and the entire college landscape is changed. I think that will be great. We should put the decision in the hands of students and families. But that’s also a huge mind-shift. That takes time to do. It can’t be mass produced.

Here’s what happens when things are mass produced in ed tech:

1) The technology makes inequities worse. We know that access is the key to good ed tech. But we come up with great products that are still stuck in a business strategy that will eventually mean students will be priced out. Again. So then we create a new problem of trying to get whatever it is in the hands of the people who need it while people who didn’t need it are further out-pacing everyone. The pandemic seemed to bring awareness to this issue by demonstrating the lack of technology in most underserved households. We also have studies that show most underserved students are taught to use technology to do rote memorization while over-resourced students use it to extend their learning. We have some good college-matching technology that is too expensive for underserved high schools to use.

2) The quality and purpose of the technology shift. I can’t tell you the number of times we’ve been approached about an acquisition from a bank that wants to use our product to refer students to student loans. Nearly all of our competitors do this. So they shift to intentionally recommending students to colleges they can’t afford because it will make them more money in a shorter amount of time. When the technology is growing so fast underneath you, it’s expensive to maintain. I often times think about this type of growth when I think about McDonald’s and that number of burgers served. The larger that number is the worse that burger is. I also think about it with tech companies like Airbnb. Think about what the houses and hosts were like before it blew up. If I make what I make and students are still informed about the best fit college, that’s a win for me.

3) We lose sight of what the end goal is supposed to be. As a teacher, I see a lot of ed tech products come through that are meant to solve problems that don’t exist or solve a problem that only exists if you use another ed tech product. For example, now everyone is spooked by ChatGPT so now I’m seeing a ton of products that detect plagiarism. Or...we can go back to having students write essays in class, on paper. I’m not sure what the big deal is. I also love ChatGPT because it makes finding things a lot easier. So why not teach kids how to find things easily and teach them how to write? I may be oversimplifying but it seems that we’re trying to make people really nervous because we’re trying to create an industry that only exists because of the first thing. Teachers have been teaching kids to write well before computers existed.

This is also not intended to dismiss all ed tech. There are lots of products that make the mundane, quick. And some of the more simple tools have been the most effective (read: that red line that appears when something is misspelled). When I ask students to research an archive, they are accessing databases all around the world. When I’m scaffolding for students, I can quickly adjust the lexile levels. I think the difference exists in what the goal is and if the point is to benefit or leverage the extremely large student market. Education is the only thing that everyone will do. So solving problems for everyone takes time even though the financial potential can be distracting.

And in order to create something for everyone, we need machine learning to scale solutions. My other big gripe is that a lot of organizations claim to support underserved students but only a few dozen to a few hundred at a time. So while we don’t want to mass produce solutions with the goal of profit-making, we do need to with the goal of reaching swaths of people. That’s slow and tailored and purposeful.

What does all of this mean for the actual machine learning? It is more financially appealing to make something that works for over-resourced students. It’s also way easier. In teacher-speak, we call this the loudest kid in the classroom. They’re raising their hands too much, loud, and interrupting everyone else. It makes it seem like they are the only ones there. And we tell them lies like, “If you have a question, everyone else probably has the same question.” They might, but chances are if you have to go out of your way to ask it, they probably don’t. And you’re wasting everyone’s time. That’s what I think about when I get the question, What are my chances of getting into [insert high brand ID college]? If that is your only concern, then this isn’t for you. And the data I’m collecting on that student and that question is distracting me from answering the questions I know the majority of students actually have. It’s the loudest kid in the classroom.

So data is collected carefully here. Models are tweaked regularly. Research is reviewed constantly. It takes time to do correctly. But again, I’m sticking to my win.

And yours. We’ve been at 100% college placement and graduation with funding for over ten years because of this approach. We take time to understand your goals and what technology can facilitate. It’s where we came up with social fit. We received questions such as where can I go to college and be safe as a Muslim student? Trans student? Where can I ensure my student receives adequate healthcare especially on a campus with high rates of sexual assault? Where can I ensure I maintain my personal values on a campus that differs from my stance? Does this change in an election year? How will I know?

Each of these examples created a new user experience and a new machine learning model, new business strategies and new data-collection methods.

The other, and most important reason, this approach is most beneficial to you is this should be a one-time deal. If we do this college-going process right, I should never hear from you again. Just like with quality furniture and handmade goods.

As a brand, this is why I wanted to Bridge to College to look less techie and more handmade. I want to remind you that things are being understood all the time, that things are a work in progress, unfinished, and unpolished.