Lessons learned from a first time founder

Avanzai is…

❌ Synthetic Financial Data Gen (April 2024)

❌ Multi-Agent Orchestration (June 2024)

❌ AI Agents for Portfolio Risk (July 2024)

❌ Workflow Automation for Financial Data (Sep 2024)

✅ AI-Powered Financial Data Analyst (Dec 2024)

1. Focusing on VCs, not on customers

I had it all mapped out in my head: Avanzai raised a $2 million in a pre-seed round backed by ABC Capital. Finally, my immigrant parents could stop asking why I quit my high-paying job as a Product Manager to become a tech entrepreneur. But after a while, I realized I was chasing VC money for its own sake, not necessarily because I needed it. In fact, the closer I got to closing a round, the less I could convincingly answer “Why do you want funding?” beyond “I don’t want to burn through all my savings.” The more time I spent on accelerator applications and VC meetings, the less time I had to talk to customers. The more impressed investors and operators were with “AI agents” and other buzzwords, I felt like the less actual users understood—or cared. They just wanted their problem solved, not to hear about fancy technology.

Once I started reaching out to users and getting genuine feedback, I realized that I should have done this from the very beginning: talking to users, getting feedback, and iterating is the fastest way to reach product-market fit — and revenue. It became clear that everything else—VCs, buzzwords, and projections—mattered far less than understanding real-world needs. For example, after talking to users, I realized that synthetic financial data generation (April 2024) and multi-agent orchestration (June 2024) were solutions looking for a problem. On the other hand, AI-powered financial data analyst (Dec 2024) was developed directly in response to user feedback, addressing their needs and delivering tangible value.

2. Focusing on the technology, not on the user

“So what do these agents do? ‘AI agents for…’?” Every time I explained what I was building, people would give me this polite nod, like they didn’t want to ask how ChatGPT factored into “agents.” Even worse, the more I went into autonomous agents, tools, etc., the more I could tell 90% of people (and users) had absolutely no idea what I was talking about. And in hindsight, I get it. I got too comfortable with the corporate, LinkedIn-style explanation of what I do—analyst, strategy, risk, management, solution scalability, multi-platform integration, adaptive workflows, (insert more jargon here). But when I pitched to actual users, they didn’t care what an AI agent was. Again, they just wanted to know if the product solved their problem.

When I started truly understanding users’ problems—to the point where they would curse about them—everything clicked. The more users hated the problem, the better I understood its urgency. For example, when we launched workflow automation, I got pats on the back from observers, but users didn’t find it particularly helpful. However, when I focused on chart generation or report generation—issues users deeply resonated with—it became a game-changer.

3. Minimum Billable Workflow

The best startups are the ones where, within 5 minutes of using the product, you immediately see why you’d pay. (Lumni and Perplexity come to mind.) Even better is when they ask you to pay, and you’re like, “Ugh, I guess?”—because you clearly see the value. The genius is that these startups know you’ll get to that realization quickly. While working on my MVP, I always wanted to figure out what the “minimum billable workflow” would be—an end-to-end user journey where the value was instantly obvious. Being anti-chat and anti-“terminal,” I knew this would be tough. Users would face a new or complex experience, or a simpler one where they might question its worth.

Having an MVP isn’t enough for pre-seed; revenue growth is what matters. And the easiest way to drive revenue growth for an early-stage startup—especially when software is no longer a barrier to entry—is making users realize why they would pay you for something they might think they can replicate with tools like ReplitAgent.

If I had gone the simpler route earlier, I might’ve launched a premium product sooner. The earlier you figure out how to compel users to pay, the sooner you can show them how bad life is without you.

4. Demo Videos Are a Good Flex, Talking to Users Is Even Better

I’ll admit it: my first demo videos took HOURS to perfect. Like a lot of AI products, the demo looked great, but actual usage left much to be desired. It’s easy to show a polished demo; it’s harder to get out there and understand your users’ needs.

Evaluation (evals) is critical for agents, especially the final 10% of accuracy, which often matters the most. A lot of my lack of posting in the past 2-3 weeks has been because I’ve been honing in on this—making the product as stable and performant as possible so it can scale easily.

As founders, we often think we know what users want, but that couldn’t be further from the truth. Founders should stay humble and listen to as many users as possible. Users provide gems of insight that eventually turn into 10x features they love and can’t live without.

5. I Am the Moat

While at Launch’s Founder U, they emphasized how critical the team is at an early-stage startup. But when the team is just you—and you layer on some imposter syndrome—you start wondering how strong that “team” really is when things get rough. We all know that building an AI startup is a race. As AI eats software, very few companies have a true technical moat.

But here’s the beautiful part: I realized I should take advantage of what I’m really good at—being myself. Cutting out the jargon when delivering what we do is my ultimate moat. I have both done enterprise sales and built / shipped products revenue generating products. I can communicate the value of what I’m building without overcomplicating it. And that, I’ve realized, is a competitive advantage very few other founders have.

So starting today, I’ll be doing exactly that. Man, 2025 will be fun...

That is all for this week folks! If you enjoyed please subscribe! 🤝