From PromptWorld to Nagent: My Journey Building the Future of Agentic AI

The plan was simple: build a store where creators could sell their prompt engineering skills. Call it PromptWorld. Watch the magic happen.

ChatGPT was exploding. Everyone was sharing "life-changing prompts" on Twitter and LinkedIn. Prompt engineering felt like the new gold rush, and I wanted to build the pickaxe store.

We moved fast. We shipped in weeks. We got noticed from around the world. I spoke to as many early users as I could, one call at a time.

But late at night, staring at our analytics dashboard and our notes from user interviews, I couldn't shake a nagging feeling. We were solving the wrong problem.

The moment everything changed

I remember the exact moment doubt crept in. A user had just bought his fifteenth prompt that week: marketing copy generators, e-mail templates, content calendars. He called me to ask if we had a way to "chain these together" so he didn't have to jump between tabs.

That's when it hit me.

People didn't want to collect prompts like trading cards. They wanted AI that actually worked for them. They wanted systems, not snippets. Solutions, not scripts.

The prompt store suddenly felt like selling individual LEGO bricks when people wanted to build castles.

The question that kept me awake

Here's the question that changed everything: what if the future of people working with AI isn't about better models or better prompts, but about better outcomes?

I started obsessing over it. I went deep into multi-agent systems, agent workflows and the patterns that were starting to emerge in how AI behaves. The more I learned, the clearer it became. The future of software wasn't apps. It was agents. Not static prompts you copy and paste. Not tools you have to babysit. Agents that think, adapt and execute, without you having to become a prompt engineer.

That realisation scared me. Because it meant starting over.

The leap into the unknown

Pivoting from PromptWorld to what would become Nagent wasn't just a product decision. It was an existential reset.

I wasn't making the leap alone. I had to convince my co-founders at the time to board what looked like a rocket ship without fuel. I was asking two talented engineering and product leaders to walk away from a working product with paying users and chase a vision that existed mostly in my head. After countless late-night discussions and whiteboard sessions, they saw what I saw: the future wasn't in selling prompts, it was in building intelligence.

Then came the harder conversation with our wider team. I had to ask everyone to work for half of what they had been drawing, and some to go without a salary for months, to keep the dream alive while we rebuilt from scratch. I'll never forget that meeting: the silence, the nervous glances, the weight of asking people to bet their livelihoods on an unproven idea. But one by one, they said yes. That's when I knew we had something special. Not just a product, but a team willing to sacrifice for a shared future.

The hardest conversation wasn't at work, though. It was with my wife, Ipsita. She had already lived through the roller coaster of my first startup. She had seen the late nights, the financial stress, the highs and the crushing lows. She had watched that first venture fail, and she had every right to be sceptical when I came home talking about "just one more idea."

"Give me one month," I said. "Just one month to prove this can work."

That one month became the company I run today. She didn't just give me time. She gave me the stability to take risks, knowing that whatever happened with the business, I had someone who believed in me even when I didn't believe in myself.

The builder years

The first version of the new vision was this: let anyone build and share AI agents, with no coding required. Not single-turn answers. Multi-step workflows. Agents that could research, analyse, create and act across different tools and models.

Here's what the entrepreneurship podcasts don't tell you: most days feel like failure. There were weeks when our retention looked like a cliff dive. There was one brutal week when almost nobody came back. I questioned everything: our positioning, our interface, even our company name. There were 3 AM rewrites of landing page copy, weekend sprints rebuilding core workflows, servers crashing during demos. The doubt was constant. What if we're too early? What if nobody actually wants to build agents?

Then people started building agents that did real work, and for a while that felt like the answer.

What the builder taught me

It took me longer than I'd like to admit to see that we had learned the PromptWorld lesson again, one level up.

People hadn't wanted to collect prompts. It turned out most of them didn't want to build agents either. They wanted the work done. The ones who did build agents ran into a second problem. Each agent sat alone in its own tab. It didn't have the brief. Nobody else on the team saw what it did. Its work had to be carried to the next person by hand, and it never added up to the number the team was measured on.

The agents were lonely. So were the people using them.

The fix wasn't a better agent. It was a team. So we left the builder category and built one.

Where we are now

Today Nagent is Multiplayer AI for Growth: people and AI agents in one shared context, delivering the number. It is three things that work as one.

• The Room. A company's people and its AI team share one room, with one thread, one board and one record. Agents hand work to each other in the open, and people approve.

• The Team. A real AI growth org, not a list of bots: the AI Chiefs of Staff, MIRA for Marketing, SERA for Sales and CREA for Content, with the team leads who do the work under them.

• The Work. The growth work, carried to the number: being found in search and answer engines, ads, social, creative, outbound, the CRM and the content all of it needs.

And at the top of the org, the newest piece: N.

Once a company has a whole AI team, the first question everyone asks is "who do I talk to?" N is the answer. N is the one teammate you talk to, by voice or by text. Behind N is your whole AI growth org. You say what you need, N hands it to the Chief of Staff who owns it, the team does the work in the Room, and it comes back to you as one card to approve. Nothing that publishes, sends or spends goes out until a person taps approve. N is in Beta.

Meet N. Speaking is working.

When I think back to that user asking me to chain fifteen prompts together, this is what he was really asking for. Not a better prompt. Not an agent he had to build. A team that does the work with him, in the open, and lets him decide.

The honest truth about entrepreneurship

Let me be real: this hasn't been a hockey-stick story. I've rewritten our pitch more times than I can count. I've had weeks when I wondered whether we should shut down and get "real jobs". I've dealt with imposter syndrome, cash-flow anxiety and the constant pressure to move faster while building something that lasts.

What keeps me going isn't the metrics. It's the people: the team that took a pay cut to keep building, the person at home who gave me one more month, and every creator who shared what our platform helped them bring to life.

I still believe the next wave of AI doesn't have to come from Silicon Valley. Nagent was founded in Bengaluru, and we build for growth teams in India and the US.

The invitation

If your marketing, sales or content team is carrying more work than it has people for, I want you to see what that looks like with a whole AI growth team beside you. Bring the number you're trying to move, and talk to N.

It was never about better prompts. It's about people and AI working as one team. And that future starts with people like you.

Pratap Behera is the CEO of Nagent, Multiplayer AI for Growth: people and AI agents in one shared context, delivering the number. nagent.ai