What I'm thinking

When you no longer read the code, you want to pick the language that runs everywhere.

Unless you’re doing something super niche, Typescript is the answer.

People will start to wake up to the fact that a monorepo is an anti-pattern for large projects.

We've had such great freedom and life improvements since switching from a giant monorepo to many pico repos 6 months ago. It's like we got rid of the chains holding us back.

General intelligence is getting cheap, but specific knowledge is only going to get more expensive.

Domain experts see things others can't, and things no model was trained on.

It's very easy for AI to lead people down a wrong path and the cost of getting it wrong is higher now.

Every second spent building somebody else's company is a second not spent building your own.

Every second spent on marketing is a second not spent building something worth talking about.

Every second spent polishing the pitch deck is a second not spent building your product.

Every second spent chasing funding is a second not spent on making revenue.

I think most people will own a business in the future.

They kinda already do. When you work for somebody else, it's like your employer is your only customer.

If you want to make more money, you need more customers or the few customers to pay you more.

When you have few customers that spend a lot, they demand more from you and their managerial role over you is stronger, so it feels like being an employee.

I’ve seen this exact failure mode kill the 3D printing industry.

I’ve worked in photolithography (semiconductor additive process), CMP (semiconductor subtractive process), and metrology (semicondutor measurement process), also at a 3D printing unicorn, and also ran my own 3D printing business.

3D printing was supposed to disrupt traditional manufacturing, but never got past prototyping, never achieved high volume, and yield is still pretty low even with the most advanced machines.

Code should be just an output of a higher level process.

That way, you can improve the process, and all the code improves as a function of it.

Instead of fixing bugs, you eliminate failure modes.

Our vision for the future is much bigger than what one git monorepo could support.

3 contrarian takes that we’re betting on for the future of AI software production:

1. A monorepo is a structural anti-pattern

Monorepos exist because coordinating across many repos was expensive when that cost was priced for human coordination. But now you can automate coordination with agents handling the cross-repo changes and pull-based flows replacing manual syncing.

I’m building the product that most teams will eventually use to exponentially improve while producing software at high scale and high yield.

Once people start noticing it is orders of magnitude better than anything else that has ever existed, we will be ready for takeoff.

3D printing never took off because it failed to solve scale and yield.

I think I know the reason why:

Semiconductor manufacturing is the most sophisticated, highest volume, and highest precision production process in the world.

My career after college started as a photolithography engineer at TI.

There, I learned how to build and monitor processes that scale, how to lead teams, and how to drive continuous improvement across cost, scrap, and yield.

Now I’m seeing the same revolution happen in software. Building, testing, and deploying software end-to-end at high scale and high precision requires the same rigor.

I don’t vibe code anymore.

I build and optimize processes that my AI workers follow to vibe code better than I would myself.

Everything is spec’ed, from how the code has to be written to every pixel you see in the UI.

Test-driven-development, RCA, validation, rollout, process control, all done by AI.

My work has shifted to finding ways of optimizing the processes, choosing trade-offs, and defining the vision.

So far, I feel like I’ve reached an exponential improvement curve.

Keep Calm. Problems are Solvable.

My first internship was very interesting.

I was a sophomore in college and worked in the Quality department at Siemens.

My main activity was helping an instructor teaching engineers Root Cause Analysis (from 5-Why to KT), risk assessment, corrective/preventive actions, etc.

I did the same training multiple times and it was very in-depth. For example, we spent a lot of time learning how to define a problem statement because everything else depends on it.

After that role, I worked in R&D, my own business, semiconductor manufacturing, startups, etc. Everywhere I worked, I was able to apply that training. I believe it's a big contributor to the success of big companies that have this knowledge: solving problems and eliminating root causes.

In the age of AI, you shouldn’t be trying to grow your thing.

You should be trying to build something that can grow on its own.

Build something only you could.

I love the work I get to do every day. It's fun!

I go to bed excited about what I'll do in the morning.

I wake up and don't want to waste any time, so I jump straight to it!

My dream is to do this for the rest of my life.

Here's my general advice for building a successful startup:

1. Don’t blindly follow anyone's advice, including this one.

2. Build a great foundation; don't build a temporary shelter, build a skyscraper.

3. Build before you sell; sales are much easier when you have a great product.

4. Build a great product; the market rewards those who create genuine value.

5. Adjust and adapt daily; this is your biggest advantage over big companies.

6. Master prioritization; your most important role should be prioritization.

7. Do things that scale; prioritize efforts that can compound and multiply impact over time.

I was born in one of the poorest areas in Brazil. My mother was pregnant with me at 14, and I was raised by my grandparents until I was seven. My mother was a housemaid.

Something that always stood out about me was a strong desire for continuous improvement and building a better life. I always strived for excellence in everything I committed to doing.