 The most sexy startup in Silicon Valley is AI startup. The hottest guy today is ML engineer. I was kidding. But most people have no clue how to build a successful AI product and how to break into AI product management. Most people like AI, but they don't know how to put the AI onto their long-term product roadmap without creating even more constraints. This is Doctrine City. I've launched an award-winning AI product as a director product. In this video, I'm gonna cover the top four most foundation knowledge of end-to-end AI product management lifecycle so that you can quickly jumpstart your AI career. Make sure to stay until the end of the video where I cover the top secret that most people don't talk about when they develop the AI product. Hey guys, this is Doctrine City, a director product featured in Forbes. I've helped hundreds of people land their dream PM job offer in fan companies and unicorn startups and continue to get promoted as a product leader. Initially, I would cover the tech trends of free product management training. Like and subscribe and check out new video every Tuesday. The four stages of AI product management lifecycle. The number one is customer discovery. This is quite similar to all the tech product. However, the product discovery in the AI process is slightly different. You must evaluate whether AI can actually solve the needs and desire and pain point of customers. Lots of you all know they just want to put AI because they're very sexy, but in reality is AI is not a good fit for several different scenarios. For example, I asked this question, what would the CEO of Delta do in the coming 10 years? One of my students inside a PMX theater, her idea was to use AI to do automated flight pass control for Delta airline. This is definitely a big no-no and very dangerous behavior because you don't want to use AI to real-time predict the flight pass or even change a flight pass because it's likely who maybe 1% or maybe less than 1% of chance AI prediction was incorrect and then you're going to have flight crashes. So what this means is when you ask any airline passengers as a customer, they never say I want AI to design my flight pass. But what they're really saying is I want to have more pleasure flight experience when I'm on Delta airline. So therefore, we must spend lots of time doing customer research and understand how exactly AI fit in and how customers is able to perceive and also use AI product. That's why all product managers need to create the end-to-end customer journey map to evaluate what does customer want and like. The second part of customer discovery is to do deep research about the technology you'll be using. For example, assuming AI is a good solution to solve the customer pain point, but you must understand the tools and models to use for AI and what kind of data collection strategy and any existing tools and models we can leverage. For example, right now, Genitive AI is very hard, but you might directly be able to use API through OpenAI, Chatcha BT and to directly build your product on top of existing model. When I build my award-winning product, which is used AI to reduce car crashes and that product received the Mayor's Best Practice Award in 2017. When we built our specific product, we didn't start anything from scratch. We look around the existing AI models that we can leverage and build on top of it. That's why we went to NVIDIA and at that time NVIDIA already had very good object detection models that we can leverage and build our models on top of it. So the different layers of models we have been using to build the specific AI product. So therefore, product managers must make technical decisions to do evaluation of existing tools and work with your engineers to understand what's the best strategy to apply AI in the initial phase of investigation. And in this video, talk about in-depth regarding how I build the AI model and to help see these reduced car crashes. You can check it out right here regarding day-in-life as AI product manager. The second phase of AI product management is to develop the AI model and collect data. The data has testing data and training data, different kind of data set, and the model itself has different kind of variables that goes into the model itself and engineers and product managers together need to decide how much data will you use to train the model and how much data is enough and where you need to consider the cost of collecting data and the cost of training your data. Because a lot of things sounds easy, but it might be very expensive to collect those data because data has a new name in the AI space is called the field of AI. Whoever controls the data is able to manage and win in this stressful competition. You also need to deeply evaluate AI governance. Now, there's a lot of ethical debate regarding when to use AI, how to use AI, and how you're able to control AI in the long run. So those weird AI ethics coming in and AI governance is quite important when we design our specific AI model. For example, the latest scandal about Sam Altman getting kicked out of open AI and invited back as I'm filming this video, there's a lot of discussion regarding why he was fired and why he was invited back. But one of the theories, which I believe the true is, he actually want to push AGI to the market sooner and making greater success and greater revenue for the company for this nonprofit. Well, of course, the board member denying and saying that it's not a reason, we never know, but this is an example of AI governance. People really have different opinion and different strategies regarding AI governance that can go way above beyond regarding a technical aspect of building the product that involves a lot of hard takes, people's opinions, society's reaction to it. So pay close attention to design the AI governance. A lot of them are, hey Nancy, how would I know what exactly technical decisions I need to make for my AI models, different variables and training data set. And I'm going to publish an AI Prime Management course to make sure it's available for everybody. And if you're interested, you can join the way this to be notified once it's available sometime in 2024. You can go to the link below and join the way list to be notified whenever it's available in 2024. The third part of AI Prime Management life cycle is developing the MVP and involve it from MVP to MMP eventually to product launch. That's a very heavy software development process and also has a lot to do with data engineers, ML engineers, they all work together to make sure the data is sufficient to train the model and also provide very good and accurate outcome. Of course, there's lots of hallucination and accuracy issues in the AI space. That's where the technical aspects are coming in. But in the product management development life cycle, the process developing the product from MVP to MMP to launch is quite similar compared with traditional product management development cycle. Just there's more involvement from the data training set and also AI ML engineers contribution. In my next video, I'm gonna dive deeper regarding all the technical aspect of AI product managers need to master without coding for them to continue to grow their career. So make sure to subscribe to channel and keep tuned when the new video is uploaded. The fourth part of AI product management is launching the product and also manage the customer adoption. When you launch any product, you need to design the go-to-market strategy. However, for AI product, not only you need to design how would you bring it to the hands of customers, but also has a lot to do with customer's perceptions and how customers are able to use the product because lots of customers like every people like us, we may not get exposure to AI. We might be scared of AI and the interaction between the human and AI might not be that easy and smooth. Therefore, we need to closely manage the user adoption. Meanwhile, we also need to manage the internal stakeholders within the company could be other traditional product managers who wasn't sure about how AI is going to impact their future job and also be other stakeholders such as legal team who have lots of concerns about AI. So many internal stakeholders you need to manage when you launch this AI product. Now here comes the most important bonus tip. Whenever we design and launch AI product, we frequently do not understand how exactly the UI of AI is going to impact the future product development cycle of AI because lots of people can create AI it's just a chat-gbt, but how you design the user interface between different users and based on specific use cases should be very customized and different based on who is the end user and what's the specific application. And for example, if you apply AI in the medical space, it will be completely different than you apply AI through Instagram online grocery shopping because in the medical field, the doctors and also nurses and have certain standards and thresholds they are willing to take AI to make decisions. On the other hand, if you want to use AI-assisted shopping bot on Instagram to buy groceries and customers probably feel more adjusted to it and get the very little negative impact just in case you bought a garlic and they give you ginger. In contrast, if you use AI chat-gbt that's available on Instagram today, probably people's adoption rate to AI and talking to AI to pick the right grocery for them is quite high because people have been exposed to those AI in power chat-gbt. If you're serious growing your AI career and try to master how to become an AI product manager and how to grow a successful AI career, make sure to join the waiters for my AI product management plus that's coming in 2024. You might be also very curious, what does it look like to become AI product manager? Is it really as sexy as what people think it should be and what exactly people do as AI product manager? Make sure to check out this video where I talk about the day in life as AI product manager, using my real life experience developing my Smart Cities product. I'm gonna see you in my next video right here. This is Doc Man City from PM Accelerator.io.