Welcome to week 1 of AI for Product Discovery & Derisking Product Bets. I'm Anne, a PM and Product Coach that's been in product for 14+ years, having worked on emerging tech products in AI and blockchain. I first started learning about AI in 2017, and I've had the opportunity to guest lecture about it at Columbia and West Point, as well as create Jumpstart Your AI Career for O'Reilly. I love to coach and teach, and I'm seeing a lot of PMs wonder how they can leverage AI to become a super IC, whether that be a senior product manager, or in product leadership but wanting to contribute more or learn how their teams can 10x themselves.
While this newsletter isn't solely about AI in product management or AI product management, this series will be focused on using AI in product discovery to derisk product bets. I've worked in four different emerging tech startups, and there are five key principles I've learned while doing discovery in a landscape that is constantly evolving, changing, but actually follows many of the same rules of product market fit.
Principal 1: You Need To Solve a Real, Painful Problem
Principal 2: Understand Your Customers Deeply
Principal 3: Show Don't Tell When Possible
Principal 4: Early Assumption Testing
Principal 5: Don't Let Bubbles Fool You Into Thinking You Have PMF
Principal 1: You Need To Solve a Real, Painful Problem
Stop learning this the hard way. I’ve worked in multiple AI startups and multiple blockchain startups as a product manager. As excited as I was to dive head first into products leveraging new technologies, some of those startups had products that, in retrospect, weren’t solving a problem that was a major pain point. They were problems that already had a lot of options, or were problems few cared about, and cared very loosely about. This is your friendly reminder, AI will not save your product, if the core use case doesn’t’ solve a real and painful problem.
Principal 2: Understand Your Customers Deeply
Not to be repetitive here, but building an AI product or jamming into your existing product won’t make customers magically appear. You still need to know who your customers are, and understand them deeply. Having worked in multiple pre-PMF emerging tech startups, I’ve seen this mistake. Yes, you can learn by building and putting your product out there, but in the paraphrased words of Shreyas Doshi, “thinking is cheap, and that’s why you should do more of it. Building is expensive.”
So the time you spent building your first failed product, you could be talking to everyone you think who would use your product and understanding them and their problems. This will test a lot of your assumptions about who your audience is, and if this even breaches the top 10 biggest headaches they have. You will experiment a bit more when you are pre-PMF, but this doesn’t mean you can avoid talking to customers, simulating their thoughts and flow, and thinking about them. You also can’t use pre-PMF as an excuse to build for “everyone.” Even with AI, building for “everyone” often is still building for no one.
When I dig into this topic more in future editions, I’m going to be sharing about how you can use AI to help you build research plans, impactful user guides for user research and testing, and doing research online to help you get to foundational understanding of who your customers are, where to find them, and what they care about.
Principal 3: Show Don't Tell When Possible
In the era of AI, you’re running out of excuses for not having anything to show around your idea. Once you have an idea of who your customer is, what their core problem is, and you’ve derisked some assumptions, ideally you’d want to rapid prototype. With tools like Replit, Bolt, Magic Patterns, and V0, there’s not really an excuse that design and engineering are too busy to help with a prototype, or that you are “too slow” in tools like Figma. AND there is still a lot of room to collaborate with design and engineering even on AI prototypes so everyone is aligned before putting things in front of a customer, and without a lot of unnecessary process.
By rapidly prototyping your ideas, you can get early feedback from customers and validate your assumptions before investing heavily in development. This helps you derisk your product bets and ensure you're building something customers truly want AND need.
In a couple weeks, I’m going to walk through how to build these prototypes, the best tools to use and why, the user guides to go with them, and how to involve design and engineering so they don’t feel like you’re going rogue.
Principal 4: Early Assumption Testing
This principal is a bit redundant, because everything I’ve mentioned up to this point is helping you test your assumptions and test them early. Here, I’m driving home that whatever you do as a PM, AI or not, you need to make sure you are testing your assumptions as early as possible. Marty Cagan talks about this a lot in his talks and books. I realize there are some organizations that make this hard, but do the best you can to test your assumptions before you have design creating designs, and engineering already building foundations for something you don’t have any early data if it will really succeed beyond some hopeful financial projections.
Principal 5: Beware False PMF Signals From Market Bubbles
As a former student of the school of “learning things the hard way,” I urge you to stop learning things the hard way, when you can learn things in an easier, less painful way. One of the early stage blockchain startups I worked at mistook a lot of the activity in the crypto, NFT, and blockchain market as signal for promising PMF. But once the market crashed around the FTX debacle, the business development pipeline dried up pretty quickly, and those outside the space were didn’t see the value in a blockchain version of a product that already works extremely well for them. While it gave me more experience in the blockchain space, the story of my product experience there was less straight forward, despite what I was able to accomplish in that scenario.
So whether you’re already building in one of these emerging tech startups, or you’re trying to land your first AI PM job, make sure you’re considering if your company is solving problems that will still be there a year or two from now.
If you’re still reading, know that I will be sharing more about specific tools, how to use them aligned to these principals and to the overall cycle of product discovery. Whether you’re pretty advanced in using these tools, or feel super behind, I’m here to show you how to think about these as tools for specific jobs, the way you’d think about deciding which kitchen utensils or appliances to use in specific scenarios. And like kitchen utensils and appliances, there is overlap in what they can do, so sometimes, it will be up to you to decide if you want to use the air fryer vs the oven.
See you after Christmas, where we will talk about using AI for market research.
Stay curious.
Anne
P.S.: I coach PMs how to 10x themselves using AI in the highest leverage ways. If you’re interested in working together, please reach out.
P.S.S.: Also, the fifth rendition of Jumpstart Your AI Career on Pearson/O’Reilly will be February 24. I will let you all know when registration opens up. Excited to teach this class again.
Share this post