--- title: "Before Echelon Philippines 2026: What I Think Abstruck Is Becoming" description: "The night before Echelon Philippines 2026, Abstruck founder Jansen reflects on building EdTech from the Philippines, what Abstruck is becoming, and the questions he is bringing with him." url: "https://www.abstruck.space/blog/echelon-philippines-2026-building-abstruck" published: "2026-08-24" updated: "2026-08-24" author: "Jansen" category: "Company" tags: ["Echelon Philippines","Student Founder","Philippine EdTech","Abstruck"] --- Tomorrow morning, I’m going to SMX Aura for Echelon Philippines 2026. I’ve been looking forward to it. I’m also nervous. There will be founders there who have spent years building companies, investors who evaluate startups professionally, operators who have hired teams and scaled businesses, and people who have accumulated the kind of experience I simply cannot have at 18. I can watch hundreds of hours of founder interviews. I can study YC companies, read about fundraising, follow product decisions, build software, break infrastructure, rebuild it, and spend an unreasonable amount of time trying to understand why some startups work and others do not. But studying startups from the outside is different from operating one for years. I know that. And I think accepting that is probably more useful than trying to disguise it. I don’t need to walk into Echelon pretending I have the experience of someone who has been doing this for ten years. I need to understand my own problem clearly, explain what I genuinely know, admit what I don’t know, listen carefully, and ask questions that are worth answering. That is a big part of why I’m going. I also want to write this down before I go, partly because I want a record of what I believed about Abstruck before those conversations have a chance to change it. ## Abstruck started with the material. I’m becoming more interested in the learner. Abstruck began with a fairly practical problem. Students have lectures, recordings, PDFs, presentations, notes, and other material scattered everywhere. Turning all of that into something usable takes work. So I started building a workspace around that process: transcripts, summaries, notes, flashcards, study chat, and ways to bring learning material together instead of constantly moving between separate tools. That is still part of Abstruck. It works today, and I think it remains useful. But the more I build, the less interested I am in defining the company by those features. Summaries are becoming common. Flashcards are common. AI chat is common. Transcription is common. Putting another AI button beside a document is not, by itself, the education problem I want to spend years thinking about. The question that has started bothering me is much more basic: **Why can two students receive exactly the same lesson and walk away understanding completely different amounts of it?** One student already knows the prerequisite concepts. Another doesn’t. One can follow the vocabulary immediately. Another spends half the lecture trying to understand a term that everyone else seems to know. One learns quickly from listening. Another needs to read it slowly afterward. Someone misses part of the class. Someone loses concentration. Someone understands the idea but cannot recall it the next day. Someone is studying at midnight with an exam in ten hours. Teachers cannot realistically reconstruct every explanation, at every pace, for every individual student during every lesson. I’m increasingly interested in what happens **after information has been delivered**. That is where I think Abstruck may have a more meaningful role: not as a replacement for teachers, but as an adaptation layer between information being given to a student and that information actually becoming understood. ## From generating study material to finding out what you actually know The product direction I’m exploring now looks less like a collection of generators and more like a learning loop: **Information → Understand → Explain → Recall → Identify Gaps → Adapt** Some of that exists today. Much of it does not. The distinction matters. Today, Abstruck can already help students work through source material in different forms: reading a transcript, reviewing a summary, writing notes, creating flashcards, asking questions, or going back to the original material. What I’m exploring next is what happens when the workspace begins asking something of the student too. One concept I’ve been calling **Test My Knowledge** starts with a blank writing surface rather than a multiple-choice quiz. Abstruck asks a broad question. The source disappears. No hints. No answer choices. You write everything you can genuinely retrieve. Then, only after you finish, the system evaluates what appears solid, what is incomplete, what you missed, whether there are possible misconceptions, and what might be worth revisiting next. The next question could then probe one of those weaknesses. I like this direction because recognition and recall are not the same thing. Seeing four options and recognizing the right answer can feel very different from reconstructing an idea without assistance. This is not something I want to oversell as a proven Abstruck learning outcome. It isn’t. But the broader idea is grounded in a substantial body of research on retrieval practice. Reviews and classroom studies have repeatedly found that retrieving learned information can improve later retention compared with simply restudying it, although the effect depends on factors such as the material, feedback, test format, and learning conditions. That makes it interesting enough to explore carefully. Another idea is **Teach Yourself**. The inspiration is simple: try explaining something in plain language and notice where your explanation stops making sense. Instead of an AI voice agent immediately teaching the student, I want to invert the relationship. The student teaches. The agent is the annoying curious learner asking: “Why?” “What does that mean?” “Can you give me an example?” “How is that different from this other concept?” The useful moment is when the student realizes, halfway through answering, that they cannot actually explain the thing they thought they understood. Research on self-explanation and learning by teaching gives me reason to take this direction seriously, while also reminding me that the design details matter and that “explain it to an AI” is not automatically an effective learning intervention. Meta-analytic work has found benefits from self-explanation prompts and from some forms of preparing to teach and teaching, with meaningful variation across conditions. Those are the kinds of distinctions I want Abstruck to respect. I’m interested in education technology, not just educational-looking software. ## Why Echelon matters to me Echelon Philippines 2026 takes place on August 25–26 at SMX Aura in Taguig. It is part of e27’s Echelon ecosystem and is organized in the Philippines by Brainsparks. The organizers project more than 3,000 RSVPs, including roughly 1,500 startups and SMEs, 1,000 corporates, and 500 investors. Those are audience projections rather than confirmed attendance figures, but it still means a very different room from the ones I usually learn in. The program itself touches several questions I’m already dealing with. There is a panel called **“Who Actually Gets Funded Here, and Why: The Funding Bias Debate,”** with people associated with Kaya Founders, Founders Launchpad, Gobi Partners, Openspace Capital, and Gentree Fund. I’m curious about that not because I expect to walk around asking investors for money, but because I want to understand what evidence makes an early company become worth paying attention to in the first place. There is another session, **“ROI or RIP: How to Measure Whether Your AI Spend Is Working.”** That sounds less glamorous, but it is close to problems I deal with constantly. AI infrastructure has real costs. Transcription has costs. Document processing has costs. LLM calls have costs. Usage has to be metered. A student product can be technically impressive and still be a terrible business if the economics do not work. And 917Ventures is running a discussion around using enterprises as real environments for startup validation: what makes pilots work, fail, or turn into commercial opportunities. Those questions matter if Abstruck eventually works with universities or education organizations instead of remaining purely direct-to-student. The wider Philippine startup ecosystem is developing, but I don’t think it helps anyone to pretend the constraints have disappeared. Government strategy documents themselves identify gaps around access to capital, entrepreneurial development, regulatory support, and digital and innovation infrastructure. StartupBlink’s 2026 index placed the Philippines 63rd globally, up one position nationally from 2025, with Manila ranked 121st among cities. For someone my age, events like Echelon create something that is difficult to reproduce by staying online: access to accumulated experience. Someone who has spent ten years operating companies may say one thing in a fifteen-minute conversation that makes me realize I have been thinking about a problem incorrectly for months. That is valuable. ## The questions I’m bringing with me I have more questions about Abstruck now than I did when it was a simpler product. I want to understand when a study tool becomes an actual learning system rather than a useful collection of features. If I claim Abstruck helps students learn, what evidence should eventually support that claim? What behavior should I measure? Convenience? Return usage? Recall? Improvement over time? Something else? How much personalization is useful before it removes the productive difficulty that learning sometimes requires? How should software support teachers without quietly drifting toward the assumption that the software should replace them? There are business questions too. Should a product for students grow directly through students? Through creators and communities? Through schools and universities? Through partnerships? Probably some combination, but I don’t yet know which deserves the most attention. And as AI capabilities become easier for competitors to reproduce, what actually becomes defensible? My current hypothesis is that the answer may be less about possessing a particular AI feature and more about the experience built around learning: continuity, knowing where you stopped, reducing cognitive load, helping you identify what deserves attention, making active study easier to start, and adapting to the fact that sometimes the person opening Abstruck is tired, behind, overwhelmed, or cramming. That is still a hypothesis. I want people to challenge it. I want someone to tell me what I’m over-engineering. I want someone to notice the obvious assumption I’ve somehow missed. I want educators and learning-science people to tell me where my product instincts conflict with what we actually know about learning. I want experienced founders to tell me which problems I am underestimating. And I want to understand what Abstruck would have to become before serious fundraising would even make sense, rather than treating investment itself as the milestone. ## A timestamp before tomorrow Being a young founder is strange because the advantage and disadvantage are closely related. I have less experience. I also have fewer assumptions about how things are supposed to be done. Most of what I know technically came from being curious enough to keep moving between layers: software, infrastructure, cloud systems, databases, APIs, AI, computing fundamentals, and whatever else I needed to understand to make the next thing work. I’m comfortable figuring things out. But I don’t want to build an education company with only a developer’s view of education. I’m a student now. I also hope to become an educator someday. The more seriously I take that, the less comfortable I am with reducing learning to “generate better notes with AI.” There is a much larger problem here, and I am still learning how to describe it properly. I don’t expect Echelon to give me the answers. I may come home without an investor, partnership, or dramatic startup story. That would be fine. If I come home with a few assumptions broken, better questions, several people worth learning from, and a clearer idea of what Abstruck needs to become next, that would already make the day valuable. If you’re attending Echelon Philippines 2026 and you work around education, learning science, student products, technology, startups, investing, or building companies in the Philippines, I’d genuinely like to meet you. **Jansen** Founder, Abstruck **Explore Abstruck · Connect with Jansen**