Five key takeaways about AI product management
The past year in technology has been marked by revolutionary advances in the use of artificial intelligence, particularly generative AI, large language models (LLMs), and natural language processing (NLP).
The generative AI gold rush has not only seen the enhancement of existing products, but also the widespread proliferation of new applications that leverage text, image, video, and audio generation capabilities.
Besides its transformative effect on the technology industry at large, these developments have also started to reshape how we think about building products and the role of product managers in this process. As AI continues to expand its footprint, it’s imperative for those of us in product management to deeply understand and leverage the possibilities and implications.
Based on my experience working on an AI products here at Intercom and recent discussions I have had with colleagues, these are my top takeaways for product managers interested in or currently working with AI.
1. Curiosity will set you apart
As the technology landscape is rapidly changing and new developments create opportunities for new applications at an unprecedented pace, the most valuable thing for a product manager operating in this space is to stay curious. Thinking critically about the space and joining the dots between all the moving parts will strengthen your product judgment – and in the age of AI, this will separate the good product managers from the great ones.
Ask yourself questions such as:
- What can this new technology do?
- How does it work?
- How does it change existing technologies and products?
- What products are being created?
- What does it enable people to do?
- How does this change users’ behaviors?
- How does this impact how I do my job?
Frequently revisiting fundamental questions like these will enable you to form opinions on how these developments shape your product, your role, the industry, and beyond.
2. AI isn’t a standalone product: your job is to understand the problem/opportunity to be solved
Alongside some really impactful applications of AI, I’ve also seen the flip-side: a rush to apply AI to products for the sake of it. Trying to shoehorn AI into a product rather than starting with the problem-to-be-solved will create products and features that don’t stick.
“Ensure that integrating AI into your product isn’t just a technological novelty but a meaningful enhancement to the product experience”
Start with your users’ jobs-to-be-done or their pain-points with your existing product. Is there an opportunity for AI to enhance/automate/transform/replace the solution? Once you’ve thought about the problems you’re currently solving or want to solve, then think more broadly to ensure you’re thinking expansively enough.
This approach requires you to understand both the capabilities of AI and the specific needs and behaviors of your users to ensure that integrating AI into your product isn’t just a technological novelty but a meaningful enhancement to the product experience. By identifying where AI can add real value and be deeply relevant for your users, you’ll avoid the pitfall of using AI as the flavor of the month.
3. The success of AI-driven products depends on users’ feelings and attitudes towards AI
If you get a signal that your AI-driven product can solve a problem for your users, there’s another thing that will determine whether or not it is successful: your users’ existing mental models of AI and their feelings around it.
If you are at the cutting edge of technology, are the majority of your users excited or hesitant to use your AI-driven product or feature? Are customers truly ready to adopt it? Or are they merely curious about it?
“Understand how your users currently think about AI”
The best way to really understand how this technology will impact your product, or what directions you could explore in the future, is to understand how your users currently think about AI. Do they see it as an opportunity or a threat? Have they even started seriously thinking about it yet?
Many of us product managers are very excited about AI and have been endlessly consuming content about it – it’s at the forefront of our minds. This may not be the case for your users – depending on your industry, there’s a chance it very much isn’t the case for them. So the question becomes how you strike the right balance between building future-facing, differentiated products while also taking your users on a journey, changing attitudes, and building new habits. What are you doing today to bridge the chasm between the now and the future?
4. Be comfortable in navigating the unknown
LLMs are black boxes. They can hallucinate, produce wildly different answers to the same question asked multiple times, reproduce biases, and are susceptible to jailbreaks. We still have little knowledge as to how they work, what they can do, and the best ways of prompting and controlling them. This makes it really hard to evaluate the performance of models and your product.
“Product managers should embrace the unknown and use it to their advantage”
Developing products in this landscape can feel shaky, but exploring uncharted territories is exciting and can pay big dividends. Product managers should embrace the unknown and use it to their advantage (we’re all in the same boat!).
Strengthen your collaboration with machine learning researchers and engineers – they are the experts in the field and can give you invaluable insights into what is possible with this new technology. Increasingly, you’ll find that you’ll have to lead with a technical/feasibility exploration before even properly defining the problem. This can feel counterintuitive to product managers, but is necessary to ensure that you don’t start from a constrained position, and have really understood the “art of the possible”.
5. Don’t get bogged down in whether AI product management is a thing or not
Lastly, there’s a lot of hype right now around AI product management that is producing an overwhelming amount of content on the subject. As ever, not everything produced will be original or useful (and that’s fine!).
I’ve seen many discussions as to whether or not AI product management is “a thing” – is it a specific role, or something that’s just becoming part of what we all do? Does it matter? Perhaps the discussion is more important than the answer itself. This space is emergent and it’s important that we have these discussions to better define and understand what it can mean for us in the long term.
For now, take it all in, be active in these discussions, seek to consolidate your learnings by applying it in your day-to-day role, and think about how it’s relevant to your product.
Stay curious :)
