Why I Built Soar

Everything about the work we do was built for a world where information was scarce. That world is gone, and the job got bigger than any of us can hold alone.

Quick note before the piece: I'm doing a live walkthrough of Soar this Saturday, July 18, at 2 PM PT / 5 PM ET. It's free, and every registrant gets the recording. Register here.

In 1986, I sat in a lab at the University of Illinois trying to teach a computer to recognize the edge of an object in a photograph. This was graduate research in computer vision, which is a corner of what we now call AI, back when almost nobody called it that. It was slow, unglamorous work. The machine was wrong most of the time. We needed faster, better computers, that didn’t come along until years later. Now facial recognition is so commonplace that we don’t even give it a second thought.

Forty years later, I spend my days sitting with high school students and their parents, talking about where they might go to college and what they might do with their lives.

Those two halves of my life had never really met. Until February, when I started building Soar.

I started building Soar just before I left on a brief vacation to the remote Osa Peninsula in Costa Rica, my second trip there. While on this trip, I spent my days thinking about what to build and why, while accompanied by a background track of the screech of macaws and the roars of the gentle howler monkeys. In retrospect, the peace and remoteness of the environment helped clarify some things that would otherwise have stayed obscure, lost in the incessant shuffle of everyday life.

Something Changed, and We All Felt It

As I thought about it, I realized that the world of college counseling had changed completely under our very eyes. Everything about the work we do was built for a world where information was scarce. Now, a student comes with knowledge they got from ChatGPT or TikTok. Much of the time, it’s false. Sometimes, the chatbot is right. Often it's half-right. Occasionally, it's right about somebody else's kid entirely, and this student has spent a week building a plan on top of it.

In the meantime, AI - just the most prominent among many other forces - is changing career paths, and the assurance that families once had - that a college degree, pursued with a decent amount of hard work, would lead to a job or graduate school, is no longer a fair assumption.

College counseling used to be mostly about getting in. Build a list, sharpen the application, write the essays, manage the deadlines. That was the work, and it was hard enough.

It isn't that anymore.

In our work at LifeLaunchr, we now get questions about whether a degree is worth what it costs. About what a sociology major actually leads to. About whether AI is going to eat a career before a fourteen-year-old finishes eighth grade. About how financial aid and scholarships really work.

Meanwhile, colleges are adding programs and majors faster than any of us can track. There are more than 1,800 four-year colleges in this country. I've visited a few dozen. I know maybe a hundred well. No counselor knows them all.

That's not a failing. It's the size of the job now. And no person holds all of it in their head, which means students get steered toward the schools their counselor happens to know, and away from the ones nobody in the room has heard of.

The Obvious Thing to Do, and Why It Doesn't Work

If families are using AI, then counselors should use AI and use it better. That was my first instinct, and it's the right instinct.

So even before I built Soar, before that trip to the tropics, I did what a software person does. I started wiring things together. Prompts, tools, data. I built skills and engineered prompts.

That's how I learned what it actually costs.

It isn't the building. It's that a homemade setup will invent a scholarship that doesn't exist, and say so warmly, and that error lands in front of a family with your name on it. It's that a good answer needs verified data underneath, not a model's memory. It's that the thing never stays built, because models change, and maintaining data is time-consuming.

And at the end of all that work, you still don’t know what your students are asking ChatGPT at midnight. And your chatbot’s advice is only as good as the profile information you give it, and that gets stale too as students grow and change.

What I Wanted Instead

I wanted a place where the student does the research, and I can see it.

That sounds small. It isn't. Almost everything that goes wrong in college planning goes wrong in the space between meetings, when a family is anxious at 10 pm and I'm asleep. If the work happens somewhere I can see, then the next time we sit down, I already know what they found, what worried them, and what they got wrong. I don't spend the first hour catching up. I start where the real conversation begins.

Building it required work I had fortunately done before. Thinking through software architecture, security, and privacy. Asking and answering hard questions about sign-up flows, connection rules, and tenant isolation. Building an architecture that hides the complexity and makes things seem simple.

So that's what Soar is. A place where a student researches colleges, majors, careers, and money, grounded in real data and the student’s profile. Where a parent can see the work instead of asking how it's going and hearing "it's fine." Where the counselor sees all of it and walks into the session already knowing.

And where, when a student asks it to write their essay, it says no and redirects them to a structured process that’s based on thoughtful self-reflection, not mad libs based on a template downloaded from a book of essays.

What It Doesn't Do

At the core of Soar is the idea that the work of a college admissions counselor has changed, but in a way you wouldn’t think of if you just heard “AI.” What it means is that with the right tools, you can focus your work on the judgment, the hard conversation, the moment you tell a family something they don't want to hear, the knowing when a kid is saying "engineering" because they mean it and when they're saying it because mom is in the room. That's the job. That was always the job. No software does that.

What the machine can do is hold the vast, unglamorous territory that no human can: eighteen hundred colleges, thousands of scholarships, the aid formulas, what a major actually leads to. It can do the part that was never the point.

No software can tell a student what story to tell on their application, but it can help you track a set of exercises and steps that will help you teach a student the art of conversational writing and effective storytelling.

Where This Stands

Soar has been in private beta since April. Counselors have been using it with real students, and now it's a HECA Business Partner. It's been invite-only during the beta, and we're opening it up shortly.

It's built for independent educational consultants and school counselors and for the families they work with. If you're a counselor, you can start for free with three students. If you're a parent, this is something your counselor brings you into.

I'm walking through the whole thing live this Saturday, July 18, at 2 PM PT / 5 PM ET. Real scenarios, real questions, no canned demo. Register here, and you'll get the recording either way.

As it turns out, when I first launched Soar, that was also on a trip: this time to frozen Montreal in the spring, where I did my work at a cafe where the owner played great music and sang along (one day singing La Vie en Rose in French to accompany the incomparable Louis Armstrong singing it in English). Travel must bring out some good ideas.

Forty years ago, I was trying to teach a machine to find the edge of an object. It says something about the malleability of careers and lives that I have become an IEC myself and am going back to software in that role. It’s my hope that Soar will help IECs all over focus not just on the human parts of their work, but also help us see each of our students as the beautiful human beings they are.

Want to try Soar? Come look at withsoar.ai.