Product Management in the age of GenAI

TLDR: Don't panic. Embrace and navigate duality. Sharpen and showcase judgement. Master prototyping. Choose the right abstractions for automation.

Recently, I've had a few conversations with colleagues about my best practices for Product Management. I've shared a brief slide deck with them and many others that outline my take on how to do it right. I've applied the approach at very different companies and for outcomes ranging from 0->1 product launches, strategic resets and platform/digital transformation.

My colleagues and others in my network have encouraged me to flesh out the slides into longer form content. I've been reluctant because I think there is already too much noise on the subject. The advent of GenAI has added to the noise to the point that I share concerns about the loss of signal. So, while I still think droning on about it at book length is not for me, I do think a few core ideas are worth an article.

Lets start with GenAI - simply put, I'm a fan. So, this post is not motivated by fear, although the image is an amusing capture of our collective subconscious with regard to the potential for AI disruption.

However, good product managers must learn to harness the obvious productivity benefits that come with GenAI. Clarity on when and how to use it to boost effectiveness comes from understanding where it doesn't belong. Therein lies the hope for all us denizens of meatspace :)

OK, so, what's the good news? What will keep us product management exemplars in business? Three key things:

1) Thriving on Contradiction: At its core, good product management is about being comfortable with contradictions...and the ambiguity it produces. What do I mean? Contradictions are tradeoffs that don't have a precise or one-off paths to resolution. They can't be put to bed once - they need constant monitoring. A few examples: a) you're the CEO of the product, but do you get to act like one? Nope! The buck stops with you but you can't exactly exert position power, b) you've got deep customer empathy, but does that mean they are always right? How do you know when to act vs tactfully mollify? c) What are the implicit assumptions baked into your product and how do they align with product or market invariants? Perhaps they are aligned with an obsolete or fading instance of a market invariant?

I think resolving or managing contradiction is what good product managers are indispensable for and that bringing AI into the mix will breed confusion, or worse, complacence. Don't go there.

2) Precision Questioning: Ok, this one is easy, particularly given its rediscovery in the form of prompt engineering. It's always been a key skill of good to great product managers. However, just like with prompt engineering, balancing the scope of the ask against the capabilities of the team is critical. Can GenAI do as good a job? Perhaps...but it will be challenged when it comes to judging the elasticity of a team's capabilities or managing product-market fit curveballs. It will also be less than adept at crafting questions that manage sensitivities arising from personal or organizational dynamics. For a particularly weasel-y take on politically attuned (or malleable) precision questioning, here's a clip. Here's what it might look like if we let GenAI lead precision questioning in an environment teeming with sensitive dynamics.

3) Cat Herding: Great product managers excel at this, particularly because they are able to do this without breaking a sweat or being seen to do it. The hard reality is that aligning the organization is a big determinant of whether a product will be successful or even see the light of day.

Here's why: with regard to commitment to the product, stakeholder engagement resembles a pyramid with architects and engineers at the bottom and the C-suite at the top. This is natural. As you go up the stakeholder pyramid, each layer is less dedicated to any one product because there are other products to consider. At the very top, unless you're at a pre-seed startup, there are other products to consider or product pivots to assess, to say nothing of financial and operational priorities that keep the business viable.

Interestingly, when it comes to influence, the same stakeholder stack looks like an hour-glass. The very top layer and the very bottom layer are primary determinants in product momentum towards market entry or acceleration. If the C-suite "decides to go in another direction", there is often little the other layers can or will do. Similarly, if there is significant brain drain, the C-suite can rarely hire their way out of the crisis.

Great product managers are able to see these shifts coming because they can look around corners while cat herding!

With all that said, getting back to the premise of the article, what should product managers do to harness GenAI? Briefly:

A) Prototyping - If you're not technical enough to use co-pilot or other GenAI variants to build prototypes of what you want the dev team to do - get moving and learn fast! If you don't, you are more likely to be left behind by those who can prototype. In some cases, low/no code tools will be sufficient. In other cases, they will just slow you down. While this advice seems biased towards software products, I'd argue that simulating a future state applies generally, and successful product managers will be excellent at creating seed implementations.

B) Generating Requirements - GenAI is particularly good at providing snippets of information from a large corpus of data about a particular topic. Use that to create requirements for well understood primitives. So, you probably shouldn't write requirements for, say, a shopping basket, from scratch. You should customize it for your needs. The same could be said for anything that has a knowledge base, documentation or a users manual. GenAI can give you a high-quality starting point, modeled on something that is close enough to your product, which you can then customize.

Remember, do not use GenAI for requirements that aren't low enough in the abstraction stack to be sufficiently close to well-known primitives. For those higher level requirements, you'll have to rely on your core skills.

C) Summarizing Feedback - Successful product management is all about having your fingers on the pulse of feedback from the market, customers, management, peers, teams, colleagues, etc. - and knowing how to adjust your internal weights to optimize outcomes. However, you can get help (because there are only so many hours in a day) from GenAI by having it summarize all the stuff that was entered, with good intentions, into various tools - Salesforce, JIRA, Confluence, Slack, ServiceNow, ProductBoard, etc. etc. Remember to have use your trusted network of colleagues to validate the summaries.

Well, that's it - if you've read this far - thanks! Comment away!

?