Hey, I’m Vinícius.
For the past two years, I was the marketing guy behind an AI platform that grew from 0 to 2 million users in Brazil.
The platform is called Tess AI. In 2024, it was ranked the #6 Best AI Software in the world by G2. That was a huge milestone for us as a team.
I learned a ton during this journey. I also got some things wrong. This article is about both.
I’ll talk about pricing strategy, product-market fit, ICP definition, product marketing, and some reflections on what I’d think about differently if I were starting today.
Let’s go.
How It All Started
Before Tess AI, I spent 7 years at Pareto, one of the leading martechs in Brazil.
We were really good at one thing: bringing data and tools together in one place. Google Ads, Google Analytics, Meta Ads, and many other platforms. All connected to help our clients get better results from their marketing investments.
We also had solid experience with machine learning and traditional AI in our projects.
Then, in early 2023, generative AI exploded. ChatGPT changed everything.
So we applied our DNA to generative AI. That’s how Tess AI was born.
The idea was simple. One subscription. 200+ AI models for text, image, audio, and video. All in one secure platform.
Think of it as the Netflix of AIs. You could start a chat with Gemini, switch to Claude, generate images with another model. All in the same conversation, using the same context.
And here’s something that made us proud. Before ChatGPT launched GPTs (their version of custom AI agents), we had already shipped something even more ambitious: the Tess Marketplace, where users could create, share, and even monetize their own AI agents.
We even joked internally that ChatGPT was copying us, given how similar the launches were. We were honestly surprised.
And we did all of this starting as a side project inside Pareto, with a small team of around 15 people, before Tess AI started running independently as a separate company.
Users Started Coming. But Were They the Right Ones?
I built a strong growth engine. Especially blending paid media and SEO, and it worked. Users were coming to the platform fast. Really fast.
We reached 1 million users in the first year. The charts looked great. Signups were climbing every week.
But something felt off.
Conversion to paid was low. And our API costs were high. Every time a user interacted with an AI model, we were paying the provider. So free users weren’t just unconverted. They were expensive.
The honest truth? We weren’t talking to customers at that point. We relied heavily on competitor benchmarking to guide our decisions.
And many of the users coming in were completely outside our Ideal Customer Profile. Teenagers making birthday cards. Casual personal use. People who would never pay.
Have you ever celebrated growth numbers that were actually hiding a problem?
That was us.
The Pricing Shift That Changed Everything
Our original model was a 7-day free trial. No credit card required. Just a usage limit for free users.
The problem: some users outside our ICP were spending AI credits on personal, low-intent tasks. We were growing in users but bleeding in costs.
So we made a change. We moved to a paid-upfront model with a 7-day refund guarantee.
And that’s when everything shifted.
Only serious users signed up. Conversion improved dramatically. We scaled revenue more than 10x in a short period.
With more predictability on the incremental MRR for every dollar spent, we could finally invest more in the team, in technology, and in marketing.
There was a bonus we didn’t anticipate. Paying users give you better feedback. Way better.
After the pricing shift, we started hearing about bugs and missing features that free users never mentioned.
We could finally understand what paying users needed and why churning users were leaving. Both signals helped us shape the product roadmap.
For example, after the shift, paying users told us the platform was too slow (a server issue we hadn’t prioritized), that they wanted to upload files by simply dragging and dropping into the chat, and that choosing between dozens of AI models was confusing. Each of these led to real product improvements, including building our own model routing system that automatically picks the best AI for each message. None of this was on our radar before.
Free users give you feedback, but it’s hard to separate signal from noise. Paying users cut through that. They tell you what’s broken, what’s missing, and what keeps them around.
Pricing was the single biggest growth lever we had. And it was the one we almost kept ignoring.
Finding the Real Customer
Negative Persona: The Users We Had to Let Go
Here’s the moment that changed my perspective.
I was reading cancellation tickets. Many users were saying the same thing: “The product is too hard to use.”
We had done everything to make it intuitive. But the platform had a lot of advanced features, and less tech-savvy users were getting lost and giving up at the first difficulty.
That’s when I realized: we were trying to serve everyone. And by doing that, we weren’t serving anyone well enough.
So I made a tough call. I defined a negative persona (the less tech-savvy user) and decided to stop chasing that audience.
It felt risky. But when I talked individually with the Product, Sales, and Customer Success leads, they all agreed. They had the same pain. They just hadn’t had the clarity to act on it yet.
The impact was immediate. Marketing became sharper. Launches became more focused. Prioritization discussions became way more productive.
“200 AIs for the Price of One? No way. That’s a Scam.”
Now, a different kind of challenge.
People on social media were commenting on our ads saying things like “200 AIs for the price of one? That’s a scam.” They thought it was a shared account scheme. Some even thought we were stealing credit card data.
We had significant investments in social media ads, so there were a lot of comments, all visible to every potential customer.
My first instinct could have been to delete them, complain, or fire back with a defensive response. But I chose a different path: education.
I started dedicating time to personally reply to each comment with a thoughtful, detailed explanation. How the platform worked, how we connected to providers like OpenAI and Google through APIs, and how the business model made it possible. Once I found the right approach, I created complete educational templates so our CS team could keep that same quality and tone at scale.
Why did this matter so much? Think about your own behavior. When you see an ad on Instagram from a company you don’t know, what’s the first thing you do? You go to the comment section. You want to see what other people are saying.
And when you find the company there, taking the time to educate and explain with real, human answers, something shifts. You start to trust them. You see they actually care.
That approach turned our comment section from a liability into one of our best trust-building channels.
When was the last time you turned a customer objection into content?
What I’d Think About Differently Today
I’m proud of what we built. Truly. But with the benefit of hindsight, there are a few things I’d prioritize earlier if I were starting fresh.
Talk to customers from day one. Call them. Run win-loss interviews. Understand why customers really choose your platform, and why they don’t. Send them messages on Instagram. Get in touch directly. As a marketer, make it your priority to understand the customer’s mind instead of relying only on competitor benchmarking, or on what your CEO, Product, or CS teams hear in their own conversations. Those inputs are valuable, but nothing replaces direct contact with your clients.
Start with a paid trial immediately. Yes, you’ll have way fewer conversions compared to a free trial. That can feel scary. But the users who do convert will give you something much more valuable: confidence. You’ll know they represent real demand. You’ll trust the data. And you’ll be able to scale your investments without guessing.
Think deeply about specificity. The AI market today is incredibly competitive. Some of the best entrepreneurs in the world are building in this space. One thing I’ve been reflecting on a lot is how focusing on a very specific problem, and integrating deeply into the workflows people already use, might be one of the fastest paths to product-market fit. It’s not the only path. But in a market this crowded, specificity can be a real advantage worth exploring early.
Invest in product marketing early. This is a personal one. When we started Tess AI, I had some knowledge in product marketing, but not enough to do a great job from the beginning. I was learning on the go: studying positioning, messaging, launch strategy. Over time I got better, and at one point we even brought the marketing team inside the product team for a few months, which made a huge difference. But that learning curve cost us time and money. If I were starting again, I’d hire a product marketer full-time from day one, or at least bring in a consultant who can own positioning and messaging without the bias that founders naturally carry.
That’s It
Building amazing SaaS is easy today with AI. The models are powerful. The tools are accessible.
Building something people actually want to pay for? That’s the real challenge.
And honestly, this is the same challenge every founder has ever faced. But today, with how easy it is to build something using AI, it’s even more tempting to fall in love with what you’re creating before checking if anyone actually needs it. Don’t let the ease of building fool you into skipping the hard part.
After some years building an AI SaaS, I recently made a career move. I’m now working at WeTransact, a GTM Marketplace platform from France that helps other SaaS companies get listed on the Microsoft Marketplace, earn their incentives, and sell to customers all around the world. New company, new challenges. And I hope to bring everything I learned at Tess AI into this new chapter.
Thanks for reading. If you want to talk more about product marketing, AI, or anything else, find me on LinkedIn. Just search for Vinícius Chacon.
It’ll be a pleasure to connect with you.
