Satwik grew up wanting to build things. That was why he chose mechanical engineering in the first place.

But when he started working with software, he realized he enjoyed a different kind of development cycle.

"In mechanical engineering, seeing an idea become something usable could take weeks or months. With software, the feedback loop is much faster. I could build something, ship it, and see customers using it. I realized that pace suited me better.”

He started teaching himself software while studying mechanical engineering. Then he found a research internship at IIT Hyderabad where he was applying AI to material science problems.

At the time, there was not much of a playbook for it. Very few professors were working at the intersection of the two fields, and even the PhD and research scholars around him were figuring things out as they went.

"It was very new. Very few professors were working on it, and there was very little research. Even experienced people didn't know that much about it yet. So I was constantly trying to find new things, learn about them, and implement them.”

For the first time, he was working on problems where nobody around him necessarily knew the answer.

"That internship gave me a foundation in understanding the latest AI, algorithms, and technologies. It taught me how to approach a problem and figure out how to solve it.”

Satwik already had a placement offer from a large company through campus recruitment. But he was looking at startups because he knew the learning curve would be steeper there than at a big company where he would be one of hundreds of new hires.

His friend Likith was already interning at Jurin AI. He reached out to Satwik and explained what they were building.

"I was completely mind-blown. I immediately looked up the company online but, to be honest, there wasn't much at the time."

His first interview was with Farz. His second was with Rise, and it ran for over an hour. They talked about how Jurin AI had started, how a simple project had grown into a product and then a company, and what it was like choosing a startup over a large multinational company.

"After that discussion, I was very interested in joining, both because of the product vision and because of the things I would be working on.”

Rise asked if he could start the next day. Then he asked if he could start right now. It was 5pm for Satwik.

"I said, yeah, let's go. I can start now. He created the accounts and everything, and I started that same day.”

One of his first tasks was integrating a customer's system into Jurin. The engineers who normally handled that work were on vacation. Satwik had never done anything like it before, and he had three days.

So he started by learning two systems at once. He had to understand how the customer's system worked, then understand enough of Jurin AI's existing architecture to connect the two. He worked on it with Likith.

They finished at around 4 or 5 in the morning.

Rise had a customer demo half an hour later.

"We barely made the deadline.”

It was a fairly immediate introduction to the pace of the company. And Satwik found the speed he was looking for.

"The speed of development is super high, and that's the main differentiator. It isn't about relaxing and trying out new things. It's more like, we have something we need to do, we need to complete it as soon as possible, and it needs to work.”

Satwik later went on to lead one of the product teams, working on some of the systems at the foundation of Jurin AI's agents.

Satwik describes two of the main systems he worked on as foundation layers for the agents on Jurin AI's platform.

The first is knowledge. An enterprise might have thousands or tens of thousands of documents containing the information an agent needs to answer a customer's question. Those documents can come in different formats and structures. The knowledge system has to make that information usable by the agent and retrieve the right answer within seconds.

The second is tools, which is the part Satwik is most proud of.

"There are many chatbots that can look through information and give you an answer. But the real change with our agents is their ability to take actions. Whether you're an e-commerce platform, in real estate, or something else, you can create a tool and let the agent perform that action. It gives customers the freedom to choose what they want their agents to be capable of."

One gives the agent access to what it needs to know. The other gives it a way to do something with that information.

Getting the first part right can be harder than it sounds.

A spreadsheet, for example, looks structured. But the information inside often isn't. Satwik found this frequently in documents used by traditional enterprise customers.

"I have seen so many customers use spreadsheets, and even though a spreadsheet is a structured document, the actual information inside is often not structured at all. The agent needs to understand that unstructured information and map it to the real answer."

The team taught the system to recognize the difference automatically and process each case accordingly, significantly improving its accuracy.

When Satwik first joined, someone else would usually decide the architectural approach.

He would receive a task, work through how he planned to implement it, get the approach approved by his team lead, and then start building.

When he became a team lead himself, the responsibility flipped.

"Instead of implementing a feature under the team lead's direction, I had to find the right approach myself, discuss it with the team, split the work between us, and make sure everyone understood the approach and didn't have any blockers.”

That also meant thinking about more than whether the code worked.

He believes that a feature is only successful when customers actually use it, can understand it without sitting through an hour long explanation, and end up spending less time on the work it was designed to help with.

"If something is successful, customers use it on a daily basis. It should be useful, easy to use, and they shouldn't face too many bugs. They should be able to understand it without our FDE or customer team explaining it in a one-hour meeting. It should actually save them time or make them more efficient so they can focus on other tasks.”

The next step Satwik wanted to make was from understanding how something should be built to understanding why it should be built in the first place.

He wanted to get closer to the customer problems behind the feature requests.

"I would like to learn more about the customer pain points and the business value. I want to be able to suggest what would actually be useful to customers and what would make a good product, instead of only implementing whatever I get from the product team."

When asked what he would build for himself using the same technology, his answer was simple. He wanted an agent that could look at all the messages coming in throughout the day, figure out what actually mattered, and turn them into a prioritized task list.

"I just want to focus on completing my tasks in priority order."

And eventually, he would like the line between enterprise software and that kind of personal agent to disappear.

"I would want everyone to be able to use it in their personal life as well and make that more efficient. You shouldn't have to use completely separate things for your work life and your personal life."

Satwik has a particular way of describing what it feels like to work at Jurin AI.

Imagine being dropped into a swimming pool before you really know how to swim.

"You're always in that position where you're not completely comfortable, but you're also not completely stressed because you don't understand anything. You're somewhere in the middle, trying to float, trying to grow, and trying to complete your tasks."

Eventually, you get better at it.

Then the pool gets bigger.

"It's a constant cycle. You start feeling a little comfortable with your current tasks, and then your responsibility increases. Staying in that middle ground becomes a normal day for you. That's how you actually grow."

Satwik wishes someone had explained that to him when he first joined.

In those first few weeks, a difficult task with a tight deadline could make him wonder whether he was actually going to be able to finish it. He worried about what people would think if he couldn't.

The customer integration was one of those moments. He didn't know the system. The people who normally worked on it were away. The demo was coming. At 4am, he and Likith were still working. And they got it done.

Over time, those situations stopped feeling unusual. The uncertainty was still there, but Satwik got better at working inside it because of the freedom he has over how he works.

Satwik is based in India and manages his own schedule. There have been nights when he worked straight through because something needed to get done, and other times when he could step away when he needed to.

Jurin AI is his first company, so this is the only way of working he has really known.

"Because this is my first company and I've worked this way from the beginning, I've become very particular about it. I think I might have a hard time finding this kind of culture again."

Which brings him back to the swimming pool.