Bryan Wan

My Ambition, for all the Physical Engineers

@CNC_Kitchen Damn. What about mattermanifest.com only distributing .gcodes to prevent stl theft

Oh, I know one of that parts. Should be mine 😅

@spi0003 Wow! Would love to see more. Do you write your own slicer or does it work with existing OrcaSlicer? What is your STL > Gcode process?

@bryanwon23 able to print up to 70degree overhangs w/o support x.com/spi0003/status…

5軸 ✖ 太ノズル ✖ 一筆書き、で陶器を色々作成しています 5軸をやってる人やセラミック３Dプリンターをやってる人はごく小数居るものの、5軸のセラミック３Dプリンターをやってる人はほぼ世界初？なんじゃないかと思います ブースではこの捻れた花瓶の作成動画をフルで流すので見に来てください

2026 is the year you realize all of insane advantages you have Airbnb coded their own payments Facebook hosted their own servers Amazon shipped their own packages Google wrote their own browser Microsoft maintained their own language Apple built their own computers

Power of Implicit CAD - Browser based, Meshless Solvers, better LLM understanding + Fully Editable. V.06 Still needs a lot of work but available to test. Taking all feature requests! LessCAD.COM

@ryusukeo1 Understood. Do you use physics simulation much? Like FEA? Do you find that solidworks is sufficient for those cases?

@bryanwon23 コメントありがとうございます。SolidWorksのボトルネックは部品点数が多くなると急に動作が遅くなることだと思います。最近使用したツールはmisumiのRAPiD Designです。

Solidworksを習得できたのは、産業機器メーカーで働いていたころ、3Dに熟練した派遣社員の方から教えてもらっていたからでした。複雑な鋳物形状や板金の逃げ、配管、シェル、分解図、歯車、取説用の分解図...などなんでも描ける人でした。今では独学でたくさんのyoutube動画から習得できると思うのですが、当時（2010年頃）はWebで調べるという発想もなく、調べる前に何度も聞きに言っていたので少し迷惑だったかもしれません。その派遣の方を心の中で3Dマスターと呼んでましたね。このポストをお読みの方の職場にもそんな3Dマスター、いらっしゃいますでしょうか？ あの派遣社員の方、元気かな...また縁があったら仕事を依頼したいな

@alexembodied @oprydai Would you be open to connecting? I’d like to learn more about the hardware process for robotics. Will be shadowing a few machinists in the coming weeks as well

@bryanwon23 @oprydai For sensor field of view studies, it can just be done in CAD by modeling the view cones and just playing with the angles to figure out stuff like blind spots. More complex sims can be done with Ansys Speos or Zemax but it’s better handled in late stage or by suppliers

how to approach a robotics startup as a solopreneur don’t try to build a “robot company.” build a narrow system that solves one real problem. what to focus on: • pick a painful niche → warehouses, farms, inspection. repetitive, costly tasks where automation actually pays. • start with software + simulation → validate perception, control, and logic before touching hardware. faster loops, cheaper mistakes. • buy, don’t build hardware → off-the-shelf arms, mobile bases, sensors. your edge is integration, not reinventing motors. • teleop first → humans in the loop. collect data, understand edge cases, then automate gradually. • data pipeline → logging, labeling, replay. robots improve through data, not just code. • service over product → sell outcomes, not machines. “we pick items for you,” not “we sell robots.” • iterate on-site → deploy early, learn from real environments, fix what breaks. constraints: • capital is tight • iteration is slow • hardware will fail so keep the system small, focused, and revenue-driven from day one. as a solo founder, your advantage isn’t scale. it’s speed of learning and tight feedback loops.

@alexembodied @oprydai What kind of software does this? I had thought simulation in terms of FEA which probably is overkill at early stage

@bryanwon23 @oprydai Simulation can be very useful if scoped properly. For instance, putting a sensor field of view cone in CAD and checking the position is relatively easy and can save a lot of time. Building your first functional prototype with whatever means necessary should still be priority

@yashhq_22 Having customers send problems directly at you and ducking VCs

What’s the hardest part of building in public? - criticism - consistency - not seeing results - knowing what to post

@catboosted Should rebrand to In hindsight partners

Girl comes up to me telling me she's an "investor" Works at Insight Partners

@sporadica All the glass half full people spend their time building, Glass half empty people end up voicing their opinions everywhere

i miss the peak days of e/acc where it used to be cool to be optimistic and hopeful about technology and the future now all anybody knows how to do is just ragebait, seethe, and lie

@jameygannon So AI native products will also die cause they are tokens dependent

we’re in the golden age of AI tools rn you’re out of your mind if you think the average pro subscription will stay at $20, $50, even $200 / mo print your money while you can

@BLUECOW009 I like to include lies in my prompt to sanity check if they catch it

I work with AI and do research but i cant shake out this feeling that the AI is lying to me and us in general, and its true we might never solve the hallucinations nobody knows how to solve it, this worries me about the future

@zeroxjackson Mac Stack

they were sold out of mac minis…

@astro_reid @NASA @NASAArtemis Amazing outlook, despite Outlook

There are no words.

@Simon_Ingari You also have an incoming buyout offer because of AI

In the corporate world, if you have: - a decent salary, - a manager who trusts you without micromanaging, - the flexibility of a hybrid work setup, - the freedom to take time off when needed, - opportunities for growth and skill development, - a supportive and inclusive company culture, You’re already among the top 1% of professionals enjoying a truly balanced and fulfilling work life.

Why is AI-native better than AI-hybrid? You are locking into a dependency on models, token costs and if you remove the AI the product stops working. Why do people think that better? Truly want to know