By Alexa Amundson, Founder of BlackRoad OS
March 2026
I got a 290 on a standardized test without doing the homework.
The kid who did every assignment, followed every instruction, color-coded her notes, and studied every night? She got a 295.
Five points. The difference between obsessive preparation and showing up and understanding.
Most of the world rewards the 295 kids. The system loves them. They follow instructions. They meet deadlines. They do what they're told, how they're told, when they're told to do it. The system was built for them.
The 290 kids are a problem. They skip assignments but ace tests. They daydream in class but understand the material. They don't do the homework but they know the subject. The system doesn't know what to do with them.
So it penalizes them. Participation grade: 0. Homework grade: 0. Final grade: B+ instead of A. Despite understanding the material just as well as the kid who got the A.
This isn't a sob story. I'm not bitter. The 295 kid worked hard and deserved her grade.
But I am interested in the gap — the space between "understanding the thing" and "performing the rituals that prove you understood the thing." Because that gap explains a lot about what's wrong with technology, education, and the AI industry.
Homework was invented in 1905 by an Italian teacher named Roberto Nevilis. Legend has it he assigned homework as punishment.
120 years later, the average American high school student spends 6.8 hours per week on homework. 13 years of school × 36 weeks × 6.8 hours = 3,182 hours of homework in a lifetime. That's 133 full days. Four and a half months of your life doing homework.
The research on homework effectiveness is... not great. Duke University's 2006 meta-analysis found that homework has no measurable benefit in elementary school and marginal benefit in middle school. In high school, the benefit exists but diminishes sharply after 90 minutes per night.
Most homework isn't designed for learning. It's designed for compliance. "Do twenty of these problems so I know you did twenty of these problems." Not because the twentieth problem teaches anything the fifth didn't — but because the volume proves effort.
The 290 kid understands this instinctively. The fifth problem was enough. The next fifteen are performance. And the 290 kid has better things to do with those hours.
The AI industry is a homework industrial complex.
The homework: Fundraising decks, pitch competitions, accelerator applications, advisory board formation, enterprise sales playbooks, SOC 2 compliance, multi-year roadmaps, competitive analysis spreadsheets.
The actual learning: Can you build the thing? Does it work? Do people want it?
I skipped the homework. I didn't do the accelerator. I didn't build the pitch deck first. I didn't hire a CTO before writing code. I didn't raise a seed round before having a product. I didn't follow the playbook.
I built 17 products on Raspberry Pis in my living room. They work. 27 agents with persistent memory, a blockchain layer, a token economy, and a mathematical framework. $150/month.
The 295 founders followed every instruction. They raised their pre-seed. They hired their founding team. They built their pitch deck. They attended the YC dinners. They're still in stealth mode with a landing page and a waitlist.
I shipped.
The 290 club isn't about being lazy. It's about efficiency of understanding.
The 290 kid doesn't skip homework because she doesn't care. She skips it because she already understands the material and the homework doesn't teach her anything new. Her time is better spent on the thing she doesn't understand yet.
Applied to building a company:
Skip the rituals that don't build the product. I didn't create a Notion workspace with seventeen project boards before writing code. I wrote code.
Skip the credentials that don't prove capability. I don't have a CS degree. I have 17 working products. Which one demonstrates more?
Skip the fundraising that doesn't enable building. $150/month buys enough infrastructure for everything I need. Why would I spend three months fundraising for money I don't need?
Do the work that teaches you something. Every product I built taught me something I didn't know. Every deployment taught me about infrastructure. Every bug taught me about architecture. The homework that matters is the work itself.
BlackRoad OS doesn't follow the playbook. It doesn't do the homework. And it works.
No framework. Every other product in our space is built on React, Next.js, or some other framework that adds complexity in exchange for convention. BlackRoad's products are vanilla HTML, CSS, and JavaScript. Cloudflare Workers with D1 databases. No build step. No framework. Just code.
Why? Because the framework is homework. It's process that doesn't add understanding. I understand the web platform. I don't need a framework to prove it.
No microservices. The conventional wisdom says you need microservices for a platform with 17 products. We use Workers — each product is a single JavaScript file. No Kubernetes. No Docker orchestration. No service mesh.
Why? Because the architecture serves the product, not the other way around. When the product is a Worker, the architecture is a Worker.
No team (yet). The playbook says you need a team before you build. A CTO, a designer, a product manager, at minimum. I have 27 AI agents.
This is either insane or ahead of its time. I'm betting on ahead of its time.
No fundraising (yet). The playbook says you need money before you build. I need $150/month and a year of 14-hour days. I had both.
I want to be honest about the risk: sometimes the homework matters.
The homework I skipped in school? Some of it would have helped. Practice reinforces understanding. Discipline builds capacity. Not every assignment was busywork — some of it was genuine practice that I would have benefited from.
The startup homework I skipped? Some of it matters. SOC 2 compliance matters when you have enterprise customers. A founding team matters when the work exceeds one person's capacity. Fundraising matters when infrastructure costs scale beyond $150/month.
The 290 philosophy isn't "never do the homework." It's "do the homework that teaches you something and skip the homework that's just performance."
The trick is knowing the difference. And the 290 kid is really good at knowing the difference.
If you're a 290 kid — if you understand things faster than the system expects and get penalized for not performing the rituals — BlackRoad is for you.
Not because we celebrate shortcuts. Because we celebrate understanding. The product works not because we followed a playbook, but because we understood the problem deeply enough to solve it without one.
The 295 kids will build great companies. They'll follow the playbook, check every box, and produce solid, predictable results. The world needs them.
But the next paradigm shift? The thing that changes how humans relate to AI? That's coming from someone who got a 290 without doing the homework and thought: if I can understand this without the ritual, maybe I can build something without the ritual too.
Five Raspberry Pis. One living room. Zero homework.
Let's ride.
BlackRoad OS — built by a 290 kid who understood the assignment.
os.blackroad.io
Remember the Road. Pave Tomorrow.