Two major developments in the last week signal how radically AI will reshape product management in the next few years. OpenAI's o3 model and NVIDIA's new Jetson Orin Nano Super aren't just incremental improvements. They're transformational tools that will redefine how we prototype, validate, and scale our products.
Let's start with OpenAI's o3 model. Beyond the buzz on X (formerly Twitter), what caught my attention is its 71.7% score on SWE-Bench Verified - significantly outperforming Claude 3.5's 49%. For product managers, this means...
Let’s start with OpenAI's new o3 model. If you're on X, formerly Twitter, you might have seen a lot of discussion and hot takes about this model. People are saying it could make software engineering nearly obsolete within six months to three years depending on who you listen to. While I can't predict the future, I believe this model will significantly impact software development and product management. o3 scored 71.7% on the SWE-Bench Verified. That means, it got 71.7% of the tasks out of the 500 issues that are a part of the SWE-Bench Verified evaluation framework. For reference, Claude 3.5 is at 49%.
With scores like that, I can only imagine how much AI will handle many, if not most of the tasks in Product Management as well. I imagine tasks like writing PRDs, writing tickets, will need a dedicated Product person supervising them, and will be overseen by product-minded people in other roles. I’m not saying the core of the Product role is PRDs and tickets, but in many companies, things like that take up a good chunk of your time if you are an IC. The gift here, is that if you as a Product can work on your craft to make product succeed, and regardless of your title in 10 years, you could still be doing product focused work in ways you could have only imagined previously. This is something I’m excited to start playing with when it’s available, because I think of the speed at which I can test my assumptions and help my customers. The key here being being good enough at the key parts of product craft to make the product successful. There are less excuses as to why it can’t succeed at this point beyond our own craft as Product people.
The second exciting development is the Jetson Orin Nano Super from NVIDIA, courtesy of “Santa” Jensen. This super computer is great for prototyping generative AI applications that need to exist in the physical world, like facial recognition or counting people entering a venue. The Jetson Orin Nano Super is powerful for handling the compute needed for building applications such as generative AI chatbots, computer vision, or voice-controlled systems. It’s a Linux machine, so you'll need some command line knowledge, but tools like ChatGPT can help you with the command line if you're not familiar.
Why is this exciting? Five years ago, I could not have imagined as a product manager I could build GPU intensive applications and prototypes from a computer I could affordably own ($249, but it’s bring your own monitor) in my own home. I’m not specifically planning on building applications that need this many resources or have applications in the physical world as of now, but if I did want to experiment, the financial bar and access bar is incredibly low now. If you have an idea, and you’ve done your thinking and research, there is less and less to block you.
These developments enable faster prototyping without significant investment or reliance on specific disciplines. Over time, we’ll see more people with deep skills in one area but also wildly proficient in others. Clair Vo, CPO of LaunchDarkly, talks about this a lot on Twitter if you want more on this. For example, software engineers will need skills in product and design too. Or a person with strong product sense who can get pretty far using AI for design and coding, though may need someone more technical to get them over the more technical challenges. The future will require balancing expertise across teams, and even PM roles will blend more into technical and design aspects than they do today.
I'm excited to experiment with these tools, and excited how they empower us to prototype quickly and efficiently. Of course after we’ve thought about the problem and if the solutions are worth prototyping; in the words of Shreyas, thinking is still cheaper than execution. I’d like to challenge you reading this to think of at least one thing worth prototyping in 2025 with the AI tools out there.
Stay curious.
Anne
P.S.: I coach PMs how to 10x themselves using AI in the highest leverage ways, while feeling more empowered and less stressed. If you’re interested in working together, please reach out.
P.S.S.: Also, the fifth live rendition of Jumpstart Your AI Career on Pearson/O’Reilly will be February 24. Register here.
Share this post