The Unfinished
I was having dinner with our new president at Maryville. Dan had just arrived, a Gen-Xer stepping into a seat a Boomer had kept warm for decades. Night and day. You could feel it immediately, the way you feel a weather change before you see it. Where the old guard saw tradition, Dan saw unfinished business. First real meal together, and I already knew this was going to be different. We were two bites into the appetizers when our waitress came over, and before she got through the specials, she stopped mid-sentence. She was looking at my lapel.
"Hey, is that the Maryville M?"
I told her it was.
"I used to go there."
Used to. Two words nobody in higher education wants to sit with. She told us her story, the way people do when they realize they're talking to someone from the place that almost changed their life. She'd been close. Really close. Six credits from a management degree. Then a child came. Then another. And somewhere between the second baby and the double shifts, Maryville became a thing she used to do. Now she was serving tables.
Six credits. She could see the finish line from where she stopped. And nobody from the institution that was six credits away from handing her a degree ever picked up the phone.
Our president listened. He didn't give her a speech. He didn't say "you should look into re-enrolling" or hand her a card for the admissions office. He pulled out his phone, wrote his cell number on the back of a napkin, slid it across the table, and said: "We're going to fix that."
She's back. Re-enrolled, closing in on those six credits, about to finish what she started years ago. Still serving tables for now, but not for long. The finish line she could see from where she stopped? She's walking toward it.
That story keeps me up at night. Not because it ended well. Because of how easily it almost didn't. If she'd been working at a different restaurant. If I hadn't been wearing the pin. If our president had been the kind of leader who says "have her call the office" instead of writing his number on a napkin. She was six credits away, and the only reason she finished is because she happened to pour water for the right people on the right night. That's not a system. That's a lottery.
She's not unusual. She's the median.
Forty-two million Americans have some college and no degree. That number is so large it stops meaning anything, so let me make it smaller. It's the entire population of California. It's more people than voted for any single presidential candidate in 2020. A nation inside a nation, and they share one thing: they started, life intervened, and the system they trusted went dark the moment they stepped away.
Higher education's response to forty-two million potential returning students has been almost nothing. A few re-enrollment campaigns. Some glossy mailers with stock photos of diverse adults smiling at laptops, the universal signal that a marketing department exists. Maybe a dedicated "adult learner office" with one advisor and a waiting list that could choke a horse. The infrastructure was never built. These people were never the customer. The admissions funnels, the financial aid timelines, the academic calendars, the advising models, all of it was designed for eighteen-year-olds moving in a straight line from high school to dorm room to diploma. Everyone else got a side door and a shrug.
I know this because I've spent years staring at the plumbing. I lead a team at Maryville that builds AI-native solutions for adult degree completion, and the deeper we dig, the more the architecture reveals itself as hostile by design. Not malicious. Worse. Indifferent. Indifference at scale is its own kind of cruelty.
Take transfer credit. A student comes back after eight years. She's got credits from a community college, a certification or two, and years of work experience that taught her more than any classroom could. She wants to finish. She's probably terrified, but she's ready.
And the first thing the system does is tell her to wait.
Wait for the transcript to arrive by mail, because apparently we still move academic records at the speed of the postal service. Wait for an evaluator to manually compare her coursework against the new institution's catalog, line by line, squinting at course descriptions and making judgment calls that vary wildly depending on who's doing the squinting. Wait while her certifications get reviewed by a different office. Wait while someone tries to figure out whether her work experience counts for anything. It usually doesn't. The system doesn't have a vocabulary for learning that happens outside a classroom. Two weeks. Four weeks. Six weeks. And during every one of those weeks, the doubt grows. Maybe this was a mistake. Maybe I'm not really college material. Maybe I should just…
That's where they lose people. Fifty-eight percent who try to reengage quit due to frustration in the process. In the waiting. In the silence between "I'm ready to come back" and "here's your plan." That gap is where forty-two million stories stalled, and nobody even measured it because the system doesn't track what it never sees.
Our waitress didn't have to wait. She had a president's cell phone number. Most people get a hold queue and a transfer to the wrong department.
So we built FeBeLabs. Not a product. Not an app. A factory for what I call revolutionware, tools that don't optimize the broken system but replace the assumptions underneath it. Everything we ship starts from the same premise: the student already did the work. The system just never learned how to read it.
KreditFlow was the first thing out the door. Not to "improve the transfer credit process." That's the language you use when you're polishing a broken system. We built it to see. To actually look at a returning student's academic and professional history and recognize it for what it is: evidence that she showed up, did the work, and got pulled away by forces the institution never bothered to understand.
When a transcript hits KreditFlow, the AI doesn't compare course titles. It reads. It looks at what was actually learned, maps it against degree requirements, forty-eight-plus years worth of data, finds the overlaps and the gaps, and produces a degree completion pathway in minutes. Not weeks. The credits that sat in a drawer for years become a foundation. The certifications become proof she already knows things the degree plan is going to ask her to learn again. The work experience becomes context that shapes the pathway instead of being ignored by it.
But credit evaluation is just the first piece of revolutionware out the door. What's behind it is what I actually care about.
Skills.
Every course a student ever took is full of skills. Not just the obvious ones, not just "can do statistics" or "understands biology." The ones hiding in the curriculum like rebar in concrete. Critical analysis. Ethical reasoning. The ability to synthesize conflicting information and produce a coherent argument by Tuesday. These skills are there. They've always been there. But nobody extracted them because the system speaks in credits and course numbers, and the world outside the system speaks in competencies and job requirements. Two languages. No translator. A student finishes Intro to Psychology and gets three credits on a transcript. An employer posts a job that needs "behavioral assessment skills" and "stakeholder communication." Same knowledge. Completely different languages. And the student sitting between them, holding a transcript in one hand and a job listing in the other, has no way to prove they're talking about the same thing.
We're building the Rosetta Stone.
Skills Intelligence takes curriculum apart piece by piece. Every course, every assignment, every learning outcome, mapped to the skills it actually develops. Not the ones the catalog says it develops. The real ones, validated against what employers hire for, cross-referenced against O*NET, Lightcast, employer-specific competency frameworks. We connect them because the student doesn't live in one taxonomy. She lives in the gap between all of them.
And a skills ontology is only as good as its definitions. Most skills definitions are garbage. Written by committee, for compliance, in language so generic it could mean anything. "Demonstrates proficiency in analytical thinking." What does that mean? To whom? A nurse analyzing patient vitals and a data analyst querying a database both use "analytical thinking" and they have almost nothing in common in practice.
So we enrich it. Subject matter experts review the mappings. Industry professionals confirm that the skills we're extracting match what they actually hire for. Then we write definitions that make sense on both sides, language a student can recognize in her own experience and language an employer can map to an open job req. We build interview questions tied to the specific skills she's developed, so when she walks into that job interview, she's not guessing what they want to hear. She knows. Because someone finally translated her transcript into the language the hiring manager speaks.
That's what FeBeLabs builds. Revolutionware. Not a chatbot bolted onto the registrar's office. AI as the substrate. The thing that sits between a student's past and her future, built to see what the old system was designed to ignore.
I think about that dinner sometimes. Our president didn't hesitate. He didn't route her through a process. He wrote his number on a napkin and said five words: We're going to fix that. And she's almost there. Six credits from done, back in the program, about to cross a finish line she's been staring at for years.
But that can't be how it works. It can't depend on a chance encounter at a restaurant, on a lapel pin and a president who writes his number on a napkin because that's who he is. Forty-two million people don't all get seated in the right section on the right night. The system has to be the thing that finds them. That sees them. That says what our president said, at scale, to every person who got close and then got pulled away.
We can't go back and call the people nobody called. But we can build something that, when the next person six credits from the finish line works up the courage to try again, doesn't make her wait. Doesn't make her prove what she's already proven. Doesn't speak to her in a language designed to keep her out.
Forty-two million people started something. The system let them down and lost their file. AI doesn't forget. It doesn't close for the summer. It doesn't have a six-week processing time. And if we build it right, it becomes the napkin. The number. The five words. At scale. For everyone.
We're going to fix that.
Building something that remembers? Let's talk or join the conversation in Discord.