AI & The Future of Product Management

AI & The Future of Product Management
I originally posted this to Linkedin in December 2022 – it caused quite a stir amongst the product management community. The comments got.. sporty.

I've been using OpenAI's ChatGPT to write the first draft of my product requirements. 20 min of work in 60 seconds. See the video below.


AI: A boon to time-poor startup PMs

let's get the obvious out of the way: this is a godsend for any overstretched PM, particularly those in startups.

80% of a PM's job is spent getting stakeholders to agree on what the requirements should be. the other 30% is writing specs, shipping, and verifying analytics. like i said: overstretched

after 10 years of writing product specs, it’s pretty easy for me to quickly form requirements in my mind. but to get the actual artifact, i have to grind away on my keyboard for 20-40 min.

with ChatGPT i spend 20-40 *seconds* writing an AI-friendly, plain-english prompt, get back human-level requirements, refactor, and move on to high value tasks like process design, strategy, and R&D.

in a few years, i believe more of a PM's clerical work will be mediated via “prompt engineering” - structuring plain-english requests to get alchemical responses from AI to create presentations, analyze data, and metabolize user research in minutes not hours.

maybe in the future, “PM” will mean *prompt manager* instead of “product manager”.

i digress..

AI: A threat to feature factories

long-term, it's hard to tell whether AI will be a boon or a threat to the craft. a lot of it depends on how the c-suite views product management's contribution to the company's value stack.

in my capitalist utopia, product managers are like mini-GMs: measured by the profit-and-loss outcomes of the products they ship. high autonomy and accountability.

however, as Marty Cagan, Teresa Torres, John Cutler, Hope Gurion and many others have stated, this is not how the majority of how PMs are managed.

the dirty open secret of our profession is that most product teams are ship-and-forget “feature factories” with most PMs operating as glorified project managers. teams are relegated to just generating product outputs based on some plain-english prompting from an executive with real responsibility for the business outcome. sound familiar?

AI: Worst Case / Best Case

honestly, i can see either scenario playing out:

best case: AI catalyzes an avalanche of innovation as it frees up PMs to focus on making more impact than just blindly shipping product.

worst case: 70-90% of PM tasks are automated, orgs would rather spend on algorithms that don’t need health insurance, obliterating demand for PMs and wiping out most of the jobs until there are only a few us left whispering to our AI coworkers the sweet nothings we learned from Reforge “prompt management” class.

The future of PM is here

either way, the future of PM work is here - it’s just not widely distributed.

AI gonna do what it’s gonna do, we don’t have much control over that.

but we do have some control in reframing how the craft of product management creates value for our employers in a world of AI.