From Chaos to Clarity: How AI is Rewriting the Playbook for Product Managers
Lessons from my conversation with ex-Google PM Assaf Reifer on building tools that tame the noise, sharpen priorities, and give PMs back their most valuable resource: focus.
When I think back on my time at Google, one of the highlights was building and scaling teams with incredibly talented product managers. Some of those PMs went on to lead big initiatives across YouTube, Google Health, and other parts of the company. A few branched out and became founders.
One of them is Assaf Reifer, a former PM on my team at YouTube in Zurich. We first met over breakfast through what I think was a LinkedIn networking experiment. He had been at Bain, was exploring his next move, and we happened to be hiring. The match worked out beautifully. He ended up becoming one of the top performers on the team and played a key role in building YouTube Analytics and the transition from the old Creator Studio into what creators now use daily.
Recently, I had the chance to catch up with Assaf on my Fireside PM podcast. He’s been experimenting with new projects, one of which could change how PMs everywhere manage the daily chaos of inputs, competing priorities, and distractions. What follows is a long, deep dive into our conversation, plus my take on what early-to-mid career PMs in Silicon Valley can learn from it.
The Setup: Why Now Is a Historic Moment for Builders
Assaf started by reflecting on what it feels like to be a builder in 2025. He’s been a software engineer, a consultant, and a PM. But he emphasized that the past two years feel different, historic even.
I remarked:
“In the last two years with advancements in AI, a lot of the knowledge necessary to build something end to end is really bridged by some of these technologies. It empowers people to realize ideas and experiments that previously required 10 people and millions of dollars.”
Think about that for a second. Not long ago, building a SaaS product that could ingest Zoom transcripts, Slack threads, and Jira tickets, then triage them into a priority list for a PM would have required a team of engineers, designers, and product folks. Now a single founder can stitch that together with off-the-shelf AI models, APIs, and some creativity.
For early-career PMs, the actionable insight is clear: don’t wait for permission to build. Even if you’re not an engineer, AI has lowered the barrier to entry so much that you can tinker, prototype, and validate ideas faster than ever. Open ChatGPT or Gemini, describe what you want to build, and let the system guide you through the concepts you don’t understand.
Assaf encourages this approach:
“The best way to start is open ChatGPT or Gemini, tell it what you want to build, and ask it how. It will respond with 30 terms you don’t understand, and you just go one by one. You ask it to explain each concept, and gradually you close the gap very quickly.”
That’s the 2025 version of “learning to code.” You don’t need to become a full-stack engineer. But you do need to become fluent in exploring, iterating, and leveraging AI as a co-pilot.
The Problem: PMs as Air Traffic Controllers
After talking about the broader builder landscape, we turned to the problem space Assaf is attacking. We discussed product managers as “air traffic controllers,” juggling multiple channels of information, each with different levels of urgency.
“Being a PM is all about prioritizing. You’re interacting with sales, engineering, customers, peers, executives. You have OKRs on one hand, and then Jira tickets or a customer threatening to churn on the other. Until recently, the best PMs just kept it all in their heads or in spreadsheets.”
Sound familiar? If you’re a PM, you’ve probably woken up to a wall of Slack notifications, 10 unread emails from sales, and a Jira dashboard full of tickets. Then, by 10am, you’re in a meeting where a senior leader asks, “What do you think about this issue that came up this morning?” And you’re embarrassed because you didn’t even know it existed.
I’ve been there. And I bet you have too.
The core challenge: noise vs. signal. PMs succeed not because they read every message but because they know which ones matter. That judgment call has historically been a mix of intuition, experience, and luck.
The Solution: Issue Center (PM Studio?)
Assaf’s project, tentatively called “Issue Center,” is a SaaS tool that ingests all the inputs PMs already swim in: Slack, Jira, Zoom transcript, and applies AI-powered rules to surface the truly critical items.
The workflow looks like this:
Integration: Connect the tool to your company’s communication stack. (His design partner is running Microsoft 365/Teams, but it could work with Slack and Google too.)
Rule Setup: Create rules that define what matters to you. For example, “API degradation impacting users” is critical. Or “customer mentions a competitor as better” is high.
AI Assistance: The system uses AI to evaluate whether inputs match your rules. It flags the items, explains why, and links you back to the source.
Prioritized Dashboard: Instead of drowning in messages, you wake up to a curated list of critical, high, and medium issues to tackle first.
Assaf demoed it live, showing how rules surfaced relevant Jira tickets, Slack threads, and transcripts. At one point, he laughed at his own naming convention:
“Clearly I’m not a marketer. It’s called Issue Center for now, but we can call it PM Studio if that makes it sound cooler.”
I told him PM Studio had a nice ring to it.
The important thing wasn’t the branding, though—it was the shift from reactive scrambling to proactive clarity.
Actionable Takeaway #1: Define Your Own Rules of Signal
Here’s where PMs can learn something even before using a tool like this. Ask yourself: What are the true signals in my work?
Is it when a customer threatens to leave?
When an API is degrading?
When an executive brings up a competitor?
Whatever they are, write them down. These are your “rules.” Even if you don’t have AI filtering your inputs yet, the discipline of defining rules forces you to separate noise from signal.
Assaf admitted that rule-writing is an art:
“The rule description is very important, because that’s what the system uses to match. If it’s too narrow, it won’t pick up. If it’s too broad, you’ll get noise. That’s why I want to make onboarding easier with quick-start templates for common rules.”
This mirrors how you should think about your own prioritization framework. If you’re too vague (“respond to all customer requests”), you’ll drown. If you’re too narrow (“only focus on API latency under 200ms”), you might miss the forest for the trees.
The Bigger Picture: Managers of PMs
Assaf also highlighted another layer of value, helping PM leads manage their teams.
“If you’re a PM lead and you have a team, you want visibility into what critical topics your PMs care about, what jeopardizes OKRs, and where they need support. This tool can give you that bird’s-eye view.”
This is huge. One of the hardest parts of managing PMs is knowing what’s actually keeping them busy. Are they firefighting customer issues? Negotiating with engineering? Or chasing shiny objects?
For managers, the actionable advice is: ask your PMs to share their “critical issue list” with you weekly. Even if you don’t have Assaf’s tool yet, that discipline will create alignment and uncover mis-prioritizations.
The Privacy Angle: Building Trust
We also talked about the obvious concern: privacy. If your tool is reading Slack messages, Zoom calls, and Jira tickets, where does that data go?
Assaf has thought about this deeply:
“This is architected as a single-tenant SaaS. It’s installed in your company’s own cloud tenant. Nothing leaves the org. Even when we use AI, it runs through your enterprise API key, which isn’t used for training.”
For PMs evaluating AI tools, this is a reminder: always ask how data is handled. At many companies, legal and IT will shut down even the coolest tool if privacy isn’t bulletproof. If you’re the PM championing adoption, anticipate those concerns and come prepared with answers.
Actionable Takeaway #2: Trust Is a Feature
In 2025, building trust is not just about having the right feature set. It’s about handling privacy, security, and reliability as first-class features.
If you’re building a product, or even advocating for one inside your company, bake trust into your pitch. Show that you’ve thought about data handling, failure modes, and user control.
Beyond Explicit Rules: The Future of Inferred Priorities
One of the fun parts of our conversation was brainstorming future features. I suggested that beyond explicit rules, the system could infer priorities by watching behavior:
If you always jump into competitor-related Slack threads, the system could propose a rule.
If you consistently respond faster to certain stakeholders, it could bump their inputs up in priority.
Assaf agreed this was interesting but also flagged the risks:
“Whenever you do something that isn’t explicitly set by the user and you get it wrong, you risk losing trust. You don’t want noise creeping into the critical bucket.”
That’s a broader lesson for PMs: don’t get seduced by complexity if it undermines trust. Sometimes a simple, transparent system is better than a magical one that feels unpredictable.
The Side Project: An AI Teddy Bear
We spent most of our time on PM Studio, but Assaf also showed me something else: a prototype for an AI-powered plush toy that serves as a conversational buddy for kids.
The idea is part educational, part entertaining. Think Teddy Ruxpin meets ChatGPT, but with parental controls and guardrails.
He tested it with his own kids, and at one point, a child said he wanted to “eat the squirrel” in a story. The system responded, “That’s not a very nice thing. Let’s try something kinder.”
That made me laugh—and also highlighted the importance of building safe AI for children.
As a parent myself, I told Assaf:
“If this thing could help kids develop critical thinking and curiosity before they jump into ChatGPT, I’d pay money for it. We don’t formally teach critical thinking to children, but a well-designed toy could do it through fun experiences.”
While this project is still early, it connects to a broader theme: AI is reshaping not just how we work, but how we learn, parent, and play.
Actionable Takeaway #3: Think About Second-Order Effects
For PMs, the teddy bear might seem irrelevant. But the lesson is this: when you build with AI, think about the second-order effects.
How does this change how people learn, not just how they work?
How does it shape what they trust, not just what they use?
How does it influence long-term skills, not just short-term productivity?
If you only optimize for immediate outcomes, you miss the deeper impact your product could have.
Practical Advice for PMs in Silicon Valley
Let’s bring this back to you, the early-to-mid career PM navigating the chaos of Silicon Valley. Here are five actionable insights from my conversation with Assaf:
Define Your Critical Rules. Don’t wait for a tool. Write down the signals that truly matter in your role and use them to triage your own work.
Build Trust Through Clarity. Whether you’re building products or pitching ideas internally, make privacy, reliability, and transparency part of your value prop.
Use AI as a Learning Co-Pilot. Open ChatGPT or Gemini and let it teach you the concepts behind the systems you want to build. Don’t be afraid of looking dumb, ask it to explain everything.
Share Priorities with Your Manager. If you manage PMs, ask for their top three critical issues weekly. If you’re managed, proactively share them. It will align expectations and reduce surprises.
Anticipate Second-Order Effects. Don’t just think about what your product does today. Think about how it changes behavior, skills, and trust over time.
Why This Matters: The Cambrian Explosion of Builders
We closed our conversation reflecting on the bigger picture. I remarked:
“You wonder if the next hundred billion dollars of market value will come not from 10 decacorns, but from a thousand smaller companies run by 5–10 people. That’s good for customers. It’s good for competition. And it’s possible because of AI.”
This is a turning point in product management. The PMs who thrive in the next decade will be those who can harness AI, not just as users, but as builders, integrators, and thinkers.
Final Thoughts
Catching up with Assaf reminded me of why I love product management. At its best, it’s about solving messy problems, shaping the future, and helping people focus on what matters most.
As you navigate your own PM career, I encourage you to experiment with AI, define your rules of signal, and always keep trust at the core of what you build.
And if you want more personalized support, I run a 1:1 executive, career, and product coaching practice at tomleungcoaching.com. If you want to try Assaf’s Issue Center tool as a design partner, feel free to contact him or hit him up on X.
OK. Enough pontificating. Let’s get back to work.
