Rebecca Ye, Co-Founder & CEO, Kyros AI

Kyros uses big data and AI to give students a personal navigation system for their education journey. Unlike a map, it evolves with you — learning about you, growing smarter over time, and guiding you on what to do, when to do it, how to do it, and with whom, rather than leaving you to work out every step from point A to point B on your own. Whether you’re 13 or 25, on the path to high school, college, or your first job, there are countless complex decisions to make, and Kyros gives you a system you can trust to guide the way.

Traditionally, families turn to school or private counsellors — often paying $25,000 or more — for help with planning: which courses to take, which activities to pursue, and which personal qualities or impact to build toward their goals. But private counselling is labor-intensive, expensive, and largely a black-box, experience-based approach. The alternative, asking a tool like ChatGPT, only works if you already know the right questions to ask. Kyros combines the strengths of both: our data delivers structured learning, while insight distilled from 1,500 professors and industry veterans adds proprietary knowledge and trends. The result is data, AI, and the know-how of 1,500 experts, all guiding you through one structured process.

Data is the foundation — without a substantial amount of it, true personalization isn’t possible. Our team includes former Googlers who’ve worked with large-scale datasets for more than 20 years, and together we’ve built what we believe is the world’s largest database in this space. We’ve analyzed information from over 1 million real applicants: which courses they chose, which activities they pursued as students, and where they ended up — the majors they picked, the careers they built, and why they succeeded. We look for patterns and work backwards from outcomes, and we also track over 40,000 scholarships and 40,000 summer programs, analyzing which students are the best fit for each.

That database is what makes deep personalization possible. We begin by understanding a student’s strengths, weaknesses, interests, values, and near- and long-term goals, so they become genuinely aware of who they are and where they’re headed. From there, we recommend the right fit. Most students can name fewer than 20 colleges, yet the U.S. alone has roughly 4,700 to 4,800, and internationally there are more than 1,500 English-language institutions. Few have a clear sense of what their ideal outcome should even look like. So we map each student from point A, their starting point, to the best-fit point B, and chart the path between them.

We work with students from 13 to 25. Beyond college admission, it matters just as much that they, first, develop genuine self-awareness, and second, build the qualities and skills needed for the AI era ahead, whether they welcome it or not. The way we develop talent has to evolve accordingly. The old model — score well, accumulate facts — no longer holds, because knowledge is now democratized and accessible at almost no cost. What matters more today is problem-solving, curiosity, critical thinking, and empathy: the skills the next generation needs to thrive in the 21st century.

When we started six years ago, our team of ex-Googlers and Stanford professors approached AI as a traditional machine-learning, neural-network recommendation engine. That changed three years ago with the emergence of large language models, and then, in early 2026, AI agents arrived in a significant way. We realized these agents could now connect the “brain” — the large language model — with real-world opportunities, and that shift pushed us into a new phase. Today we’re building agentic workflows that help students translate where they are into concrete problem-solving skills, through programs and mentorship in workflow development, robotics, and the kinds of jobs that don’t exist yet. We’re preparing them for work that hasn’t been invented.

AI can serve as an accelerator. I’m a perpetual optimist, and I believe it will bring real abundance to how we live, work, and study. That said, it’s essential to discern how you use it — to master the tool rather than be mastered by it. A student who simply asks AI to write an essay for them, without engaging their own curiosity and critical thinking, is being mastered; their reasoning atrophies. But one who uses it to ask better questions, dig deeper into a subject, and access knowledge they couldn’t previously reach is doing the opposite. Like any transformative technology, it can serve you well or badly depending on how it’s used. The judgment and guardrails around that are still maturing, but it starts with the awareness of mastering rather than being mastered.

The most obvious are what people already call AI engineers and AI verifiers — roles that simply didn’t exist before. Understanding what it means to hold one of these positions, and how AI actually works, matters enormously. Across manufacturing, logistics, retail, and other enterprise sectors, an entirely new category of AI-focused jobs is emerging: professionals who bridge powerful large language models with the CRM, ERP, accounting, and HR systems that traditional businesses already run on. Their job is modernizing those systems to be AI-enabled — distilling knowledge, skills, and tools so legacy sectors can put them to use.

Institutionalizing know-how is another recurring theme. Previously, expertise was scattered — tribal knowledge that left the organization the moment an employee did. Now companies want to distill that into a structured knowledge base, thinking in terms of semantic and ontology layers. What used to be a nice-to-have is now essential. Priorities have shifted too: where startups once chased hyper-growth, we now look for defensibility and durability.

Robotics is undergoing its own transformation. We’re talking about world models — simulating the physical world and the laws that govern it — so robots can take on tasks people don’t want to do, from dangerous work like fire rescue to everyday chores like cooking, laundry, and cleaning. That field is only just beginning; the technology isn’t there yet, so there’s still substantial mechanical-engineering and physical-modeling work ahead, and a growing need for talent. We’re building the pathway to support that development — and AI has to be in charge there too. These systems run on what’s often called a “first brain,” embedded at the edge, so they can respond to their environment in real time.

We’ve been fortunate. Six years ago, we recognized that even education — traditionally reliant on human teachers in classrooms and counselors in one-on-one conversations — needed big data and AI. Large language models and AI agents have simply accelerated that journey. Some industries will likely become obsolete, but we’re only getting started, and the pace is accelerating.

The next phase, I believe, is personalized learning, which demands not just knowledge of how to teach a subject, but a deep understanding of how each student learns. People differ: some are auditory learners, others visual; some prefer hands-on work, others thrive through collaboration. Constraints on teachers and resources have historically made that kind of individual attention impossible at scale. AI agents change that. This is where we’re investing the bulk of our research and development, working to make it a reality very soon.

It’s funny you ask — I had breakfast with our business school dean from the University of British Columbia just last week, and I’ve also stayed in touch with UBC’s engineering dean; that’s where I earned my master’s degree. Much of what we discussed that morning centered on how learning styles have changed. When I was a student, 20 years ago, the framework was rigid. The business school understood the value of problem-solving — we were embedded in projects across different industries — but that’s as far as it went at the time.

Today, we can simulate the real world with remarkable fidelity, which accelerates the learning curve considerably, and I consider that a genuine advantage. When we work with our students now, we place them in simulated environments where the stakes feel authentic but the consequences don’t follow them, because they’re still learning. That freedom lets them experiment far more — their exposure and their communication bandwidth are ten, even a hundred, times greater than what earlier generations had.