Episode 73: Vinícius Chacon - Learnings from Launching an AI Product

Speaker A: It's very easy to create amazing AIs that do everything, especially with live coding. But many of these AI solutions deliver results that customers don't actually want to use. So it's very important to really understand is there any potential customer that will be happy, uh, on using that? So this is an important question to ask before launching.

Speaker B: You are listening to this is product marketing brought to you by Products Marketing Hive, a product marketing community that gives back. I'm um, your host, Louis Se Diu. Product Marketing Hive and this podcast are supported by Product Marketing Edge, a technology product marketing consulting firm. In this episode, Vinicio Shakon, marketing lead at WeTransAct, shares his experience of launching an AI SaaS product. Let's dive into it right now. Hi Vinicio, welcome to the show.

Speaker A: Hi Louise, thanks for having me.

Speaker B: To get started, tell us a little bit about yourself and your product marketing journey.

Speaker A: As, uh, most product marketers I don't have a linear path. I started at 12 years working with design and architecture with my father. Then I started civil engineering. But I switched to industrial engineering and in college I worked in a, uh, consulting company as marketing director. Then I worked about 10 years at one startup called Pareto. It was one of uh, the Brazilian leading my tech markets in paid media management and software development there. But in 2023 at Pareto, we decided to launch an AI SaaS called Tessai. Uh, I will talk more about TESS later. But the point is, the team for TES AI were about 25 people. We were five for marketing and we had to create a brand to acquire customers and to manage every product launch. But nobody pointed to me and said, now you are a product marketer. Anyone, including me, knew about product marketing in uh, 2023 here in Brazil we just knew that uh, we needed to launch new features and new AI modules almost every day. So during the launches, uh, the product discussions, pivoting, messaging, I needed to search a lot about it. It was very interesting because during my searches I learned about product marketing and started studying about it. It was like love at first sight. I think that's my journey. Uh, working with design as a, ah, teenager, then graduating in engineering and working with marketing, product management, giving me the basis to work as a product marketer. So this was my path into product marketing.

Speaker B: It's great to hear that product marketing was a level of first sight to you. Um, that's beautiful. You've worked in tech space in Brazil. How would you describe the product marketing and startup landscape there today?

Speaker A: Okay, okay, ah, that's a good question. Because first product marketing Brazil. I see that uh, this is a new profession here. Most the founders that I talk to have no idea about what product marketing is but it's not their fault. Even most marketers don't know about product marketing. In my recent research I found that There are only 2,800PMMs in Brazil. We have more than 200 million population here in Brazil. So it's just a few product marketers right here. So probably there are many more Brazilians doing product marketing work but they don't hold the title of PMM or simply don't know about it. About the startup landscape in Brazil, I see that there are um, many startup hubs in Brazil with many founders creating great products. For example Nubank, 13 years old bank, achieved 112 million new clients in Brazil, is, is expanding in many America's countries and is, it's the largest digital bank in the world outside Asia. So it's a very big startup that came from Brazil. But we have some challenges here. Most of powerful marketing tools like Clay, HubSpot and others are often expensive for Brazilian founders and it's hard to compete with foreign salaries. So the best developers are often hired by US and or European companies. That's a uh, very big challenge for us. Uh, and so these startups need to be creative and customer centric, should be able to compete. Uh, but uh, they often grow selling just in local markets because we have a population of 200 million people. So there is a lot of market to explore here in Brazil.

Speaker B: That's very interesting, thank you for sharing. So let's talk about this recent global product marketing report you created. What I find interesting to me was the product marketing to product management ratio. So tell us a little bit about it and your opinion on the ratio, does it reflect the reality?

Speaker A: That research was very interesting because the specific charts that mentions about the product marketing to product manager ratio was the most commented on my report. Maybe everyone is trying to find the perfect ratio and I had uh, many insightful conversations with PMMs around the world about that. So let me go in depth about that data. I did that report using the LinkedIn data. So uh, there are 1 uh.3 billion users on LinkedIn and it's incredible the type of insights we can extract from there. The ratio from 1 to 16 is the number of product managers and product markets around the world. But most of the companies uh, that is in the report have a ratio about 1 to 10 or 1 to 8. So what I found is that there are companies with a high ratio Amazon, Apple and on the other side we have smaller ratios like Uber, Google and Matt. But why such a discrepancy difference, uh, uh, on those companies? I've Talked with many PMMs recently about this and S XPMM from Amazon and indirectly from Microsoft and they brought some insightful things to me. Some of those companies prefer to use outsourced PMMs. Another thing is that more technical companies like Amazon with AWS, the PMMs don't have influence in product roadmap, uh, so they are working more on just on the launch side instead of the technical uh, sides. People call them uh, core PMM and GTM pmm. So uh, in more technical companies the role of guiding the products is more on product managers reducing the needs of pmm.

Speaker B: That's interesting. Thank you for sharing the insights. Let's talk about the AI product launch. Give us a bit of context. What was the AI um product and what problem were you trying to solve?

Speaker A: It is called Tez AI. It is a platform where you can use more than 200 AI modules for text, image, audio, video, avatar. Luis, let's imagine that you are paying a uh, ChatGPT subscription. You are super happy with its capabilities. But someday Claude launched a new AI model way more powerful than Chat GPT. Everybody is talking about great things that you can do with cloth, but you can't try it because you are already paying for a ChatGPT subscription. And then um, Google came and launched VO3 for video and Nano Banana for images. Way better than ChatGPT. Again you are stuck in OpenAI subscription because you can't afford more than one subscription. Here comes Manu's AI helping you create websites, slides, agents, uh, with multi model in just one chat. Again you are stuck with OpenAI. So uh, we created Taz AI to be like the Netflix of AIs with only one subscription with a similar price from OpenAI you can use more than 200 different AI models in just one chat. Includes all that uh, was recently launched. So test value proposition is to offer professionals the use of the world's best AI models include the use of the use and creation uh, of agents with just a single subscription.

Speaker B: So in your opinion, what made this AI product meaningfully different from a traditional SaaS product launch?

Speaker A: Okay, the first one I think that is the cost of AI because when we are talking about an um, AI SaaS we have to pay the AI providers such as OpenAI and drop to Google every time a user uses the available AIs so we can't simply offer a free trial uh, because it's way more expensive to have free users on a AI SaaS compared to a traditional SaaS. So we have a margin challenge. We need to plan the pricing based on consumption and a long chat with an AI has a long context, especially if you upload documents. The longer the conversation context, the greater the number of tokens used and the higher the amount we have to paid to the AI provider. But it's not simple to explain to users that we will charge based on tokens spent instead of messages exchanged or just offer unlimited usage. Lots of customers ask us so there are uh, competitors that simply limit the context window and other underhanded things deceiving the user, uh, autosave tokens and money. But the quality of the AI responses drops significantly. So we chose not to to do that. Uh, another challenge is that the users are used to using AI for free because ChatGPT Gemini has a customized user suites. So you need to educate the marketing showing that if you want to use high quality AI models you need to pay for that. It's not simple and we need to educate the market to show that it's, it's way better to create an AI agent instead of just using a single AI model in a chat. The last challenge very specific for an AI SaaS like Tess is the broad audience. We decided something quite different that to not narrow our audience for just marketers or another specific Persona, we decided to create an AI that competes directly with chat PT Gemini Claude Manu's AI for professional usage. So that was one of the biggest challenge because on every launch we need to think about different use cases for different audience to communicate the value of that new feature. So that was uh, something very challenging for us.

Speaker B: One of the complexities was that this was a product land motion rather than sales land. So there was much less direct interaction with buyers. How did that affect the product launch?

Speaker A: That was a big challenge because at the beginning of tests instead of talking to customers, we relied just on competitors benchmarking. And this was a big mistake for us. We have listed all competitors, their features, communications and positioning. We were addicted to competitor research and we are trying to blend everything together, every great feature from our competitors together. I can't say that it will wasn't good because this helped us to have the best benchmarking sources but we are not communicating with customers so it uh, significantly delayed our growth. For example, we brought in many users who were completely outside our ideal customer profile and who were not going to convert later. But that wasn't clear to us. We Weren't talking to them. We are just seeing the number of free users increase rapidly. But with low conversion rates to paid

Speaker B: users you mentioned one of the main challenges was that you attracted a broader base of audience with strong interest in product but it wasn't working for sales. When did you realize the target needed refinement?

Speaker A: I remember when I was trying to define our Personas the most wrong thing for uh marketers that our Persona is everyone. So uh, we are trying to okay let's try to define some Personas to try to create specific messages for each one um of them. And I tried to analyze our cancellation ticked uh our churned customers. I saw many people complaining about finding the products difficult to use. Of course we did everything to make Tez AI to seem as a uh easy to use product. But the less tech save users gave up uh, uh at the first difficulty. So that's when I realized that we were trying to embrace an audience that should be a non ideal audience. The less tech savvy users uh, maybe uh it's it was about 50% of our uh audience. So in Brazil we had, we had some competitors who served this less tech save audience well uh with simpler platforms and much more focused on education than on platform quality. So we decided to stop fighting for this less tech save audience. It was the best thing we did because the marketing, launch and product roadmap discussions could be much more focused on the intermediate and advanced users whom we served very well.

Speaker B: You also made changes to pricing to help address this challenge. What did your original pricing model look like and tell us a little bit more about what wasn't working about it.

Speaker A: There were seven free days to use with a credit limit and you don't need to put your credit cards at the beginning. After that the user had the option to subscribe to the platform for $10 per month on the initial plan. So this was our first pricing model. We had many users creating free accounts using credits create images and images. Uh has a high API cost when dealing with high volume. In most cases uh, the use of images was entirely personal. For example a uh, daughter, um, birthday cards, uh, totally relevant when we are trying to increase our revenue. We quickly reached 2 million users. So you can imagine our costs of uh, AI but we had a low conversion rate and high API costs due to AI consumption. When we are using um a free trial we need to reduce the benefits of paying users because we are using parts of their ah uh money to, to pay for the free users. So it's worse even to the paying users. And the worst part, with low conversion rates for new subscribers, we couldn't scale the team couldn't scale the advertising investments or anything else, uh, and we lacked confidence in the product's potential.

Speaker B: That's interesting. So after the changes, did buyer and customer behavior shift in a noticeable way? Were there any metrics that showed the shift?

Speaker A: First when we switched uh, the price, we switched to a model where the user pay upfront, uh, for the platform, but the user could cancel and get refunds within seven days. That was when everything changed and we started to grow because all the free users that were irrelevant to us stopped converting in the platform and just relevant users keep converting the platform and we began attracting only much more qualified users who would pay to use this platform. We were able to scale our market investments, our technology team and our commitments to this um, platform. We increased sales and monthly revenue more than tenfold in a short time. It was one test AI before that and other test AI after that. Interesting. The change in the pricing which we often overlooked was perhaps the one that had the biggest impact on us. Uh, it allowed Staz AI to succeed. I think pricing for product markets is one of the most important things to pay attention when we are talking about uh, product launch.

Speaker B: How do you think pricing influences how customers perceive an AI product's value?

Speaker A: When I was analyzing the comments on our Instagram ads, I saw many people who were suspicious about how it would be possible to access 200 AIs by paying price of just one. There were lots of comments about uh, this is scam. They thought it was a shared account scheme, they thought it was a steel credit card data. Uh, seriously, uh, I read everything in the comments. Yeah, I think that the pricing has a lot of influence uh, on our product perception. But uh, in this case we instead of just uh, refuting and arguing and discussing with customers on the comments in Instagram, we used the comment section to respond to each message explaining its uh, how it actually worked. We explained about how API worked, how we were partnering with OpenAI Google to provide that. So it was something was a marketing initiative to not just try to convert that audience but try to educate the market. Because imagine our acquisition channel was from Instagram and Facebook ads. So when anyone see our ads on Instagram, for example, before clicking they went to the comment section to see what they are talking about it. But we tried to answer, to reply every comment to try to educate our customers. So when anyone go to the comment section the user could read our responses, learn how it works and have way more confidence to Convert.

Speaker B: So let's say if you were launching another AI product tomorrow, what would you do differently from day one?

Speaker A: Uh, I have three things that I would do differently from day one. First talk more with customers, call them, get feedback better monitor the use of each feature and especially conduct win loss interviews. I think that's very very valuable to understand how we are perceived and the main purchasing criteria. I think this is one of the most important to to bring uh any strategic vision to the founders, to the um product team, to the marketing team, to every everyone in the company. The second thing is considering that is a uh AI test or AI product I would offer a paid trial from the beginning or at least seven days trial but with usage limits and the registered credit cards. I think this helps from the start to discover which audience is interested in paying that just in using our product for free. It helps us scaling uh market investments and also ensures consider that the paying users will complain about every problem in your platform. So I think that bringing paying users they will bring to us valuable feedback to us to prioritize the roadmap. So this is something important to the product team, to the marketing team, to the CS team. M and the third point I would work with a more specific problem instead of competing with more generic platforms because this allows uh us uh for truly integrating AI into the customers they use especially on the platforms they already use. For example for product marketers I would create uh AI agents that monitor the message positioning and launches of competitors and bring everything to battle cards to be used uh with the sales team uh the better the adoption rate will be as it will be different from using any more generic AI to do the same job.

Speaker B: That's great lessons to share. So for product marketers who are preparing to launch an AI product now, where do you think they should spend most of their time before launch?

Speaker A: So I think that analyze competitors features, position and messaging. Yes that's super important but mainly talking to customers, understanding how they think, how they speak, what they use today because it's very easy to create amazing AIs that do everything especially with web coding. But many of these AI solutions deliver results that customers don't actually want to use. Uh instead of just doing uh an AI product that looks like something very interesting. I prefer to be very careful not to delude ourselves and create something that customers don't actually want. Uh we've seen this a lot so it's very important to really understand is there any potential customer that will be happy uh on using that. So this is uh important question to ask before launching.

Speaker B: Great thoughts. Thank you for sharing. So before we go, any final thoughts you'd like to share with the audience?

Speaker A: I think that I just wanted to thank you for the opportunity to speak with you and the PMM Hive audience. I really loved participating in the podcast. If anyone wants to Talk more about PMM, AI or anything else, just add me on LinkedIn. Just search for Vinic Chacom. It will be a pleasure to connect with you.

Speaker B: Just that great. Thank you. Thank you for joining us today.

Speaker A: Thank you for having me.

Speaker B: Luis thank you for tuning in to this is Product Marketing brought to you by Product Marketing Hive, a product marketing community that gives back. Check out our website productmarketinghive.com to join our community to meet fellow product marketers and access free resources. If you enjoyed our podcast, please subscribe and give a five star rating on the platform of your choice. See you next time.