Yesterday I came across NeoCalculator, an open-source calculator OS built by a 15-year-old from Almería named Juan Ramón. He announced a grant application, and the project has been getting attention since his first thread on X back in March. A custom CAS engine on an ESP32, a scripting language built in two days, fluid dynamics simulations, a particle sandbox, neural networks, fractals, optics, chemistry… all in eight months.
The kid is impressive. That’s not the problem. The problem is what’s happening around him.
Everyone claps. Almost nobody helps.
Open the replies to any tweet about a teenager building something. You’ll find a wall of fire emojis, “this kid is going places”, “the next Linus Torvalds”. And they mean well. Of course they do. A 15-year-old shipping code and soldering boards deserves recognition.
But recognition is not guidance. And I see a pattern I recognize painfully well, because I lived it.
Track’n’view
When I was a kid I built a project called Track’n’view. I was obsessed with it. I thought it would change everything. I poured hours, weeks, months into it. I was convinced I was sitting on something big, and the excitement was intoxicating. Every feature I added felt like proof that I was right.
But I had adults around me who knew what they were looking at. People who cared about me enough to say uncomfortable things like “this part doesn’t work yet, fix it before you add something new”, or “who is going to use this?”, or the hardest one: “maybe it’s time to stop”.
At the time, I hated it. It felt like they were killing my momentum. Like they didn’t believe in me. I remember feeling deflated, frustrated, even a bit betrayed. Why couldn’t they just be excited with me?
It took me years to understand what they gave me. They taught me to evaluate my own ideas outside my own optimism. To play devil’s advocate against myself. To distinguish between “I can build this” and “someone needs this”. That skill has saved me more times than any line of code I’ve ever written.
What I see in NeoCalculator
I ran a proper analysis on the project. Not because I enjoy tearing apart a teenager’s work, but because the claims on the website and README deserve to be taken seriously, and taking claims seriously means checking them.
The README says the CAS is implemented. But open issue #82 shows that x - 2x returns -4 because x is silently treated as 4. The comparison table against TI and NumWorks has factual errors (NumWorks listed with WiFi and a SymPy CAS, both wrong). The hardware is labeled “open” under CERN-OHL-S, but the KiCad files aren’t in the repo. A strategy document claims “12x ROI per unit” with no measurement behind it. And 13 of the apps in the launcher don’t open in the emulator.
Meanwhile, the repo has 124k lines of application code written in eight months. The commit history shows 160 commits from copilot-swe-agent and 15 from Claude. An entire scripting language was merged in two days. There are 75 documents in the docs/ folder, most of them AI-generated spec packs and ADRs for software that doesn’t work yet.
This is not a kid lacking talent. This is a kid lacking someone who tells him to stop adding features and fix the ones that are broken. Someone who says “you have three product names and zero working products” and means it with love.
The market doesn’t care about your age
The project positions itself as a competitor to TI, Casio, and NumWorks. The reality is brutal:
- Calculators are sold to exam boards, not students. Without exam approval, a graphing calculator is a toy. In Spain, graphing calculators are banned in the PAU. In France, you need certified exam mode with a blinking LED. In the US, TI owns 80%+ and Desmos is now built into the digital SAT.
- The “open source” pitch doesn’t sell calculators. NumWorks started open and locked its firmware in 2021 for exam integrity. Students didn’t leave.
- The CAS is not a differentiator. Giac (the engine NeoCalculator uses) already runs for free on Casio and NumWorks through KhiCAS.
- The €15–25 price claim is parts cost, not retail. Small-batch production with case, keypad, battery, compliance testing, and margin puts the shelf price above NumWorks’ group-buy pricing.
A 15-year-old can’t realistically navigate radio-equipment compliance testing, manufacturing contracts, or warranty obligations. And there’s nothing wrong with that. He’s fifteen. But someone around him should be saying this out loud.
The real failure is ours
I’m not writing this to dunk on Juan Ramón. He’s doing more at 15 than most engineers do in their first five years of career. The Youth Hacking 4 Freedom submission is honest about what’s unfinished. The trouble is that the website, the README, and the strategy document don’t share that honesty.
And that gap, between what the kid knows is true and what the public-facing materials claim, is where the missing adult should be standing.
The adults in Juan Ramón’s orbit seem to fall into two camps: the ones cheering from the sidelines (“amazing! keep going!”) and the ones who aren’t there at all. I don’t see the third kind: the one who sits down with him and says “let’s look at this together. What actually works? What are you claiming that you can’t back up? Where should you focus?”
That third person is the most valuable one a young builder can have. And they’re almost always missing, because being that person is uncomfortable. It feels like you’re crushing a dream. You’re not. You’re shaping it into something that can survive contact with reality.
Optimism is a fuel, but rarely a compass
When I look at NeoCalculator, I see my own younger self. The rush of building, the scope creep disguised as ambition, the conviction that if I just add one more feature it’ll all click. I see 75 planning documents for software that doesn’t work yet, and I remember writing business plans for projects that had no users.
The difference is that someone grabbed me by the shoulders and said “stop. What do you actually have? Not what you plan to have. What do you have right now?” And I had to sit with the answer, which was much less than I’d been telling myself.
That moment sucked. But it was the moment I started learning how to build things that last.
Juan Ramón has the talent, the drive, and clearly the stamina. What he seems to lack is the person who loves him enough to be honest. The one who says “you have a fantastic portfolio piece and a learning vehicle. That’s not a failure. But if you call it a market-disrupting product, you need to back that up, and right now you can’t”.
If I could tell him one thing, it would be this: the people who cheer you on are nice. The person who tells you to stop is a gift. Find that person. Listen to them even when it hurts. Especially when it hurts.
Your future self will thank them. I know mine did.
Update, June 2nd. A month after I wrote this, the exact dynamic I described played out in public. Juan Ramón’s project got featured in La Bonilista, David Bonilla’s newsletter reaching 16,000 engineers. In the thread that followed, a user made a calm, legitimate point: the economics don’t add up, because he finished his degree 20 years ago with a simpler calculator and never had to change the battery. A fair question. The kind of question a mentor would ask over coffee.
The response? He got piled on. “Thanks for coming to talk about yourself”, “You don’t seem like the type people take orders from”, etc. The actual substance of his criticism — that the cost narrative is built on a straw man — went unanswered. I replied agreeing with him: even among graphing calculators, affordable options exist. The comparison falls apart if you look at it for more than a minute.
This is the pattern. Someone raises a valid concern. The crowd treats it as an attack on the kid. The kid never gets to engage with the criticism, because the adults around him are too busy performing solidarity to let the uncomfortable question land. The critic walks away. The kid keeps building on unchecked assumptions. Everyone feels good about themselves while the wall between a talented kid and the feedback that could actually help him keeps growing taller.