I started my career around a year eight years ago in advertising technology before moving into ERP systems development for big four clients.Over the last couple of years, I've been in a product, lead product manager role focusing on delivering custom AI solutions and developing an internal product, from the ground up for enterprise level companies. Right now, I'm also mentoring small businesses on how to use tech more effectively. And today, I'm here super excited to talk with you about, what it really means for me and based on my own experiences to build an MVP for AI, specifically products, that solve problems. We all know that product managers carry lots of pressure. Definitely, that's true for me. And if a product succeeds, a lot of people would say usually it's a team win.

If it fails, whose fault that would be, Very often, you may hear that it's a product manager fault. Well, today, I want to flick the things a bit and focus on how we define success for AI products instead of how smart can we make it. I'll try to focus on how useful can we make it. On the right of your, screen within the poll section, you should see a quick question that I'd like you to answer thinking of what comes to your mind when you think of AI, MVP. Whether it's, maybe a chatbot, LLM based chatbot, maybe you imagine a quick automation, maybe it's a proof of concept. Today, I'd like to discuss this with you together and focus on a clear and value first road map.

You might have heard before and even during today's, sessions, that at this point of time, unfortunately, maybe many AI projects still fail. In my opinion, I believe that we often scope to broadly, chase the perfect model, or skip the massive work. And I believe that many of you know that it's, about cleaning the data or talking to the real users. Well, the fancy AI model is just the tip of the iceberg. Right? Beneath the surface, there is a lot of work on data cleaning, user feedback, and adoption challenges, and many of adoption challenges, I would say. If you don't nail those first, your MVP must probably sing, hopefully, no. And today, hopefully, we can talk how not to, let that happen.

Let me just, jump right into an example. I recently worked, with a mid sized company, in the computer and electronics manufacturing space on their first AI powered, support chatbot. I really appreciated about this project that the client truly approached it as an MVP or pilot. They wanted to test first the value, focusing on a single use case, and we identified that that would be a, support for the developers instead of jumping straight to the end user functionality without having a clear picture of what it's gonna bring to their business. We launched the first version in just a month, observed how test users interacted with it, then made improvements for deployment for the first version. I think that because we started small, we learned fast. And before long, the, bot wasn't just a prototype. Right? It became a trusted teammate, helping onboard users faster and making the hardware of the clients, easier to set up.

That success helped them get buy in for the next iteration, which is now tied directly to the revenue growth. Moving to the enterprise level companies. Here, it's also true that small scope doesn't mean small impact. In other house build project I worked on recently was about automating Excel based reporting. And I'm sure you know that a lot of companies, especially in the mid, even large sized companies, very often rely heavily on Excel for their operations. We combined spreadsheets and PDF into a silk single view. We used some LLMs, of course, where it made sense only, kept control with simple purpose based logic, and avoiding added features, which no one asked for yet. And the result was, around 60% of reduction manual hours and happy stockholders, making faster decision, which we proved with two pilot cases running for finance and marketing functions.

That's MVP at work. Throughout my whole career working on data driven product, in, many conversation with tech founders and product leaders, I noticed a pattern which helped me to kind of solidify what it means, for a to successfully launch an AI product. So I wrap those lessons into something simple, literally the simple framework, as you can see on the screen. And each letter stands for a principle I found essential. Starts more, in involve clients, measure impact, prioritize simplicity, leverage storytelling, and ensure quality, which is one of the most important ones for me, like, from a personal, value perspective. If you want to learn more, I wrote an article about this. And if you follow the QR code on the screen, you can learn more about this. We all love clever talk. Well, personally, I do. But what I've learned a lot is that complexity kills adoption.

Early on, I make a rule, solve 80% of the problem with 10% of the effort. Use rule based filters when it's possible it makes sense before jumping into training the models. Build flow that, feel intuitive. If your MVP feels like instead of a shortcut, you add an extra work and you need to explain all over again to your users what it does and what problem it solves, well, most probably, you've missed the point. Of course, talking about the point where business meets the attack, your MVP is a life experiment in return on investment. This is how what I see that the enterprise level companies, how they take a decision about implementing AI is when it's possible to prove that there will be a return on investment. I'm based personally in Houston, and there are a lot of companies here in enter oil and gas, energy.

And so I worked a lot with different startups that actually get into the production stage only with the return investment being proved prior to that. Look for early signals in there, time saved, cost cut, revenue potential, smarter decision, and support road being reduced. Those signals tell you whether to invest further or not. A few guardrails really quick, something that I really like to stick to. Honestly, unfortunately, I don't see them, followed enough in early stage startups or first time AI product teams. A lot of them tested AI going into production. I always, advise to stick to them, standard software development, strategies when you have the staging environment in place. This would allow you to catch issues before customers, ever see them. ML ops pipelines, make sure your experiments are repeatable and not just one ops.

Monitor for drifts because model age and data sheets even when you don't expect it. Those practices, of course, don't eliminate surprises, but they reduce the risk. And thing and, from my personal experience, things will go wrong. No road map is perfect, but having a solid risk mitigation strategy, that's what sets a successful product apart. Having the time running off, let's just have some final thoughts figured out. I'm one personally, when it comes to my approach, I I define MVP as not the final product, it's the first promise. Two basic MVP will get your product ignored, full version, sustained value. Keep it simple, keep it useful, deliver early. And as I like to say to my teams, it must be just enough for people to fall in love with it, what it does and how it does it.

That's how you build momentum and deliver the value. Well, thanks so much for everyone, for joining today. I'd love to hear your stories. Stay in touch. Keep the conversation going. You can find me on LinkedIn or catch up me after the session. If you have any questions, please drop them in chat. I will stay for a couple of, minutes more. And let's connect on LinkedIn and, keep the conversation going about how we go from good to great and how we build something that users, fall in love with. Right? Thank you. Yeah. That's a good question, Kruthika. Thank you so much. How do you know when to stop with the MVP? Well, that's a that's a hard question. And I think specifically for, product managers as what I've been mostly, practicing in.

It takes some experience, some courage to be able to face that the product and the road map is not hitting the milestones as is expected. So before I had you jump into any sort of development, you should have your major milestones defined, which would help you to sort of define whether you are on the way of meeting these milestones. And if it takes time to get to the next stage, that's a red flag, let's say, for you to consider whether you still should keep going with the next stage or no. Definitely, it's a good practice to iterate and keep going because we're working in software development. Definitely, we all practice a agile approach. I I would not want to stay very negative, but sometimes, unfortunately, not all the concepts are must go into production. MVP is just a test of a concept. So always have on your consideration, the chance that the product is not going to have a goal for it.

I hope I answered your question, and please feel free to reach out so we can chat more about this. Terry, definitely happy to connect on LinkedIn and chat more. Kathy, thanks for your question. In your experience, what is the average time for an MVP to be delivered? That's a great question. Definitely depends on the complexity of the concept and the complexity of the tag being involved in there. So if we are talking about, something that would, for example, require to gather additional training data for your model, of course, you would include that into account. If we're talking about the MVP development having all the necessary data and, miles and the road map being developed and, you have a team ready, that would usually be something like three to six months, to have something to start with. Of course, just again, comment on that. It depends on the complexity of the concept and the tag being involved, and the availability of the data. Hope that answers.

Happy to continue the conversation. If you're there, feel free to reach out. Well, thanks so much for joining. I hope you found this helpful, and I believe the recording is gonna be available for, you to, catch up with the slides. And, please feel free to connect with any questions you have. Really appreciate you sharing your stories, your experiences. I think it's really great to have this community. If, the product manager management is something that you, keeps you inspired, have can recommend you some communities, some books to read, or, things to listen. Happy to do so. So I'm personally just recently, joined, the products that count, community with the great great mentors, support available if you are on your product management journey. So would definitely recommend joining that. But please, yeah, get in touch, and, I'm sure I will find some more recommendation for you.

Thank you so much, everyone, and have a great rest of your conference and a great rest of your day.