All writing

An essay by Manthan Rajput

Building at Substrate

I didn’t really know what to expect when I started working with Substrate.

It’s one of those startups where you can open a conversation about marketing and somehow end up talking about clinical records, AI agents, biology, or why a certain product decision actually matters.

And honestly, that’s what I like about it.

Substrate is building AI systems for medicine and biology. Right now, a big part of what we’re working on is making sense of the mess that medical data can become. Prescriptions, lab reports, notes, documents, old records, everything sitting in different places.

The interesting part is that we’re not just trying to put an AI chatbot on top of it and call it a healthcare product.

There’s a lot more underneath.

I’ve been working across marketing and software, which means my days can look completely different from each other. One day I’m thinking about how we should communicate something on the internet. The next, I’m looking at the product itself and trying to understand how something works or how it could work better.

That overlap has been one of my favourite parts.

I’ve always liked building things, but working at a startup has made me realise that building isn’t just writing code.

Sometimes building is figuring out what people should see.

Sometimes it’s figuring out what shouldn’t be built.

Sometimes it’s sitting in a call and realising that the thing everyone thought was obvious actually isn’t.

And sometimes it’s just opening the project late at night and thinking, “wait, we can actually make this work.”

That last part happens more often than I’d like to admit.

What makes Substrate especially interesting to me is the problem itself. Healthcare is one of those spaces where you can’t just throw technology at a problem and hope it works. If an AI system says something, you need to know where it came from. If it doesn’t know something, that needs to be clear. And ultimately, there has to be a human making the important decisions.

Working around that has changed how I think about AI.

I used to mostly think about AI in terms of what models can do.

Now I think a lot more about what you can actually build around them.

The interface.

The data.

The workflow.

The trust.

The tiny details that nobody notices when everything works, but completely change the experience when they don’t.

And because Substrate is still early, I get to see those decisions happen instead of just reading about them later.

There’s something really cool about being in a room where the product is still becoming what it is.

You’re not just maintaining something that already exists.

You’re helping figure out what it should become.

I’m still learning a ridiculous amount every week, but that’s probably the best part.

I get to write.

I get to build.

I get to break things.

I get to fix them.

And I get to work with people who are trying to build something that could genuinely matter.

That’s a pretty good place to be right now.