Every month I teach a Tech Term You Should Know (TTYSK) and a tech essay to level up your technical literacy and collaborate well with dev teams. Ask me anything and I'll cover it in an upcoming issue.

This issue's TTYSK is "Webhook". Scroll to end to learn more 👇

I get asked this question a lot: “How do you see AI changing the PM role?”.

.. and the follow-up question: “What AI technical skills do I need to know as a PM?”

AI is a paradigm shift that we’re only beginning to wrap our heads around. There will definitely be changes to the PM role in the upcoming years, but that change will be slow to happen. I’ll explain why in a bit, but first I’ll share my predictions for how I see AI changing the PM role in the coming years.

There are two key shifts to the PM role that will happen over time:

1/ The first is a strategy shift from “feature-first” to “AI-capability” thinking

I predict that as AI features continue to become more ubiquitous, product managers will start shifting from a feature-first to AI-capability thinking.

The traditional PM “feature-first” approach is to identify a user problem, come up with specific features to solve them, then deliver detailed specifications to engineering.

This approach works because it creates clear, predictable functionality. For example, if a product manager is trying to solve for better search capability for users, they might say "Users need better search, let's add filters and sorting."

However, AI is pushing product managers to think differently. Instead of “discrete” features, AI-feature PMs have to think in terms of AI capabilities as dynamic building blocks that when used together can solve for multiple use cases at the same time and evolve over time.

This shift in thinking requires us to move from "what specific features do users need?" to "what could AI do with our data and domain knowledge to create new value?"

Let’s take the same search example and see what solutions AI-capability thinking might look like instead:

• Users can ask an AI chatbot: "Show me red dresses under $100 that would work for a summer wedding" (aka natural language queries)

• AI model can learn from user behavior: If users who search "summer dress" often click on lightweight, floral items, the AI prioritizes similar items

• AI model can personalize results by combining past purchases and browsing history to rank results

• AI has semantic understanding capabilities like understanding "evening wear" includes gowns, cocktail dresses, and formal suits

I’ll quickly sum up the differences between the two:

2/ The second key shift is evolving dynamics between cross-functional teams.

The current M.O. requires cross-functional teams to be highly collaborative and specialized with the PM at the center orchestrating everything.

AI workflow tools at some point will start to redraw theses boundaries by giving PMs the tools to take on tasks typically reserved for other roles, especially designers and engineers.

Let’s take for example, the traditional process of validating a new feature. A PM would:
→ write a product spec
→ wait for designers to create wireframes
→ collaborate with UX professionals for a prototype
→ then work with engineers to build an initial version

This process is typically non-linear with lots of back and forth and can take weeks or even months. With AI tools and workflows, this timeline will likely get compressed mostly from product managers being able to do much of the initial legwork without the communication overhead of working with another person.

These AI-powered tools really expands the capability of every product manager, but it doesn’t eliminate specialized roles.

In other words, we’re not eliminating the entire race, we’re just moving the starting line forward.

Product managers being able to do more upfront allows for more efficient collaboration, not less need for collaboration. AI tools is simply allowing each role to operate at a higher level and focus on their unique value-add rather than preliminary work.

It’s still too soon to tell how and if this increased capability for product managers will create any lasting or permanent changes to the standard way of operating for the product management industry as a whole, though I predict it will to some extent.. but it won’t be immediate or anytime soon.

The question this all raises is how should product leaders think about balancing skill developments and role boundaries: When to use AI tools for quick interactions versus when to engage a team member of a specialized role? How to use these AI tools and workflows to enhance rather than to bypass team input? Where do we draw the line between rapid prototyping and product-ready solutions?

So what AI technical skills do you need as a PM?

This requires a nuanced answer: it depends on what type of AI feature you’re working with engineers to build.

There’s a plethora of different uses cases for AI features across the board, from recommendation engines, dynamic pricing to virtual assistants, knowledge bases and even mental health bots. Each of these different features/products is built with different technologies so depending on the AI feature, you will need more specific domain technical knowledge.

However, regardless of what the AI feature is, AI is still AI. So there will be core AI skills you need to understand that will be applicable no matter what your future role or specialization is.

Here’s a list of the core AI skills you want to start brushing up to provide you the underlying context you need:

This core AI knowledge will already put you above most product managers. If you’re looking to switch into a role where your team is actively developing an AI feature. There’s a lot more to learn, but I’d start from the top to bottom of this list as it gets progressively more technical as you make your way down.

You’re not behind on AI!

Just last week I was reflecting during a conversation with a top product leader on how resistant even the top tech companies in Silicon Valley have been in adopting AI into their product management day-to-day process and product lifecycle.

But you wouldn’t be able to tell based on the sheer volume of AI content on your social media feed.

The reality on the ground is human behavior is slow to change and typically requires a strong motivator to overcome deeply rooted habits and cultural/organizational patterns. Unless you work at a company whose core product is an AI product, this will likely continue to be the case until said strong motivator comes along.

So you can rest a little easier to know that you’re not behind. If you want to get ahead of the curve to grow your career in the direction of being an AI features PM, now is a great time to start learning core AI skills I outlined and start positioning yourself as someone actively improving in the AI skillset.

👉 📬 What would you like me to cover? Ask me anything and I'll cover it in an upcoming issue.

💡 Tech Term You Should Know (TTYSK)

Webhook

A webhook is a way for one system to automatically send a real-time notification to another system when a specific event happens. It acts as a trigger that instantly pushes updates instead of waiting for the other system to check whether there’s been an update.

Think of it like package tracking. Without notifications, you’d have to keep refreshing the tracking page to see if your delivery has arrived. But with notifications, you get an alert the moment it’s at your door. Webhooks work the same way by pushing updates the moment something happens instead of making your app constantly check for updates.

Let’s say your product integrates with Stripe and a customer pays an invoice. Here’s the difference between using webhooks and not using them:

• Without Webhooks: Your app has to keep asking Stripe, “Has the invoice been paid yet?” over and over. This approach:

  • Causes delays – Customers might not see their payment reflected immediately.

  • Wastes resources – Your app keeps checking even when nothing has changed.

• With Webhooks: The moment the customer makes a payment, Stripe sends a notification to your app. Your app can instantly:

  • Mark the invoice as paid.

  • Send a confirmation email to the customer.

  • Update the user dashboard to show the payment status.