#148 Why Soft Skills Are the New Hard Skills, Navigating the AI Revolution in Product Management & Design w/ Margaret-Ann Seger

Playback speed

Share post at current time

0:00

#148 Why Soft Skills Are the New Hard Skills, Navigating the AI Revolution in Product Management & Design w/ Margaret-Ann Seger, Head of Product & Design at Statsig

#148 w/ Margaret-Ann Seger, Former Head of Product at Statsig, incoming member of the Product Staff at OpenAI

Oct 06, 2025

I used to say, “Don’t let AI do your research.” I’ve changed my mind.
That shift started before this interview—after I ran a complex API exploration through an AI research assistant and got back a thorough, sourced report with working links.
But my conversation with Margaret-Ann Seger (who leads product and design at Statsig, the intelligence infrastructure platform for feature flags, experimentation, and analytics) turned that insight into conviction. And the timing makes this even more interesting: Statsig recently announced it’s being aquired by OpenAI.
If OpenAI sees enough value to combine forces, the way Margaret and her team work is worth studying.
Here’s the core change I’m making in my own work: let AI collapse the paperwork so humans can concentrate on judgment. When I dump a messy outline into a model and it returns a clean structure in minutes, I don’t feel threatened; I feel focused.
Margaret described a world where PRDs update themselves from meeting inputs and auto-ticket the next steps. That’s not cutting corners. It’s cutting ceremony. The value we bring isn’t keystrokes—it’s synthesis.
Synthesis shows up in how we decide what to build when shipping gets cheap. AI lowers the barrier to creation, which raises the bar on taste. It’s not enough to ship more; you have to choose better—distill the real pain, reconcile what users say with what they actually do, and shape solutions that feel right in the hand.
Margaret triggered a new habit for me: I now write a one-paragraph “taste test” before we commit: Why this problem? Why now? Why this approach? If I can’t explain it plainly, we aren’t ready.
The conversation also reframed “soft skills” as the durable edge. You can’t paste three years of team history into a prompt. Reading the room, sensing when engineers don’t buy a solution, remembering why a past decision failed—these are still human advantages. Margaret called out the friction every PM knows: users tell you one thing in interviews and do the opposite in product. Someone has to hold both truths at once and decide. That someone is still us.
One practice of hers made that human edge tangible: make customer support everyone’s job. At Statsig, support isn’t a silo. They rotate it. Designers answer confused tickets and see where the UI collapses. Engineers feel the frustration firsthand and often fix root causes quickly. It’s tempting to route everything through a bot for speed, but there’s a hidden cost: you lose the raw empathy that powers taste. We’re piloting a similar rotation and tracking the fixes it sparks.
Another theme was moving learning into production. Prototypes were born when shipping was expensive. As that cost falls, high-fidelity demos give way to small, live experiments that gather real data. Margaret’s ideal cadence is to spend more time on problem analysis and then release multiple small bets behind flags. I’ve started doing the same for ambiguous flows: define two or three minimal viable variants, ship them to real segments, and time-box the learning window. Data beats debate.
On the tooling side, Margaret pushed me to point AI at sources of truth, not just the documentation. Docs always lag. Code doesn’t. Her team is exploring agents that answer questions grounded in the codebase and SDKs. I loved the example of customers repurposing Statsig’s experimentation tool to benchmark models and prompts offline—a reminder that good tools get bent into new jobs in the AI era. We’re trialing a code-aware path for technical support and an internal agent trained on our repos for integration questions.
Something else I’m now normalizing: don’t hide your AI usage. Margaret hired a PM who clearly used AI on the take-home. That wasn’t a disqualifier; the deciding factor was his judgment. The stigma needs to go. Show your work, raise the standard, and trade playbooks. We’re adding a simple line to retros: “How did AI help?” When the practice is visible, everyone gets better faster.
Two moments from the interview keep replaying for me. The first was our “soft skills” segment, because it names what PMs actually do when the tools get powerful: we arbitrate between truths, people, and paths. The second was personal and small—Margaret and her husband use AI to make songs for everyday moments and stories for their kid. It’s a reminder that this wave isn’t only about efficiency; it can unlock more human connection at scale.
Here’s where I’ve landed:
I no longer treat AI as a novelty or a threat. I treat it as an accelerant. It compresses the “what” so I can deepen the “why.”
I’m biasing toward live learning and away from document theater: fewer perfect specs, more real outcomes.
I’m putting empathy on the front line (support rotations), taste at the gate (the one-paragraph test), and code at the center of truth (repo-grounded agents).
If you’ve been skeptical like I was, start small: choose one active project, let AI handle the formatting and ticketing, and spend the saved hour with a customer or sharpening the problem statement. You may find your job doesn’t get smaller. It gets truer.
And in a world where a company like Statsig is merging with OpenAI, getting to that truer version of product work isn’t optional—it’s the edge.
Listen to The Way of Product: Apple Podcasts or Spotify

Margaret-Ann Seger, leader of product and design teams at Statsig, and after the acquisition from OpenAI, part of their Product Staff, discusses the evolving role of AI in product management on the latest episode of The Way of Product.

While AI can automate many tasks, human judgment remains crucial. She shares insights on how AI can supercharge productivity and reduce drudgery, allowing PMs to focus more on strategic thinking and deeper customer understanding.

Margaret also explores the idea that future PM tasks will blend with design and engineering roles, facilitated by AI tools. She remains optimistic about AI's impact on creativity and productivity.