Product leadership and AI strategy lessons from a career at Google, Amazon, Slack, and Glean

Executive overview

Most product leaders conflate doing well with being promoted, but the real accelerant is impact — doing the right thing for the company even when it costs you personally. Operational chaos doesn't kill companies; missing product-market fit does.

A career built on following great people and making decisions right, not making right decisions.

Two habits that drive career success

• Do a great job at the job you're actually in — not the next one
• Table stakes for PMs: technical depth, deep product knowledge, metrics fluency
• Understand people's motivations as precisely as you understand user behavior
• Ask more questions; assume you know nothing (Benioff's "beginner's mind")
• Impact is the only measure — not "did I launch" but "did people use it, did it move the company"
• Early career example: turned down her first management role because the team shouldn't have existed; her manager respected it and found her a better one

You don't need a career plan

• Anxiety about five-year plans is wasted energy — follow people, not domains
• Go where the best practitioners are; their skills will transfer to you
• Network from working together compounds over time (PayPal mafia, Google alumni)
• Advice from a mentor: "Take a job where hiring people makes their careers"
• Financial returns are the hardest thing to predict; learning is the safest bet
• If a company fails, skills don't — they travel with you

Well-run companies don't always win

• High executive turnover, constant reorgs, people yelled at — and the numbers are amazing: this happens
• The inverse: great CEO, well-oiled machine, flatlined
• What actually matters: product-market fit, distribution, a working sales team, enough runway
• No PMF is a death sentence; operational chaos during hyper-growth is normal, not fatal
• Chaos threshold shifts by stage — past ~5,000–10,000 people, professional management becomes load-bearing
• Some chaos is fine; changing strategy and direction constantly so nothing gets finished is not

What Bezos and Butterfield taught her

• Bezos always spoke last in reviews — went around the table first to hear everyone's view
• His consistency made him easy to operate under: you knew what he cared about before you walked in
• Told her "that is your advantage" when she said a competitor had 10x more engineers; pivoted her frame permanently
• Butterfield wrote a four-box master plan in 2014 — search, Slack Connect, platform, AI — and it never changed, only the annual work changed
• His core practice: "I have to feel it, I have to try it" — pushed for working prototypes, not mockups
• Prototype example: put everything in the interface behind one button to reveal what was actually essential

Building cross-functional alignment with engineering

• Finding the right engineering partner is a hiring criterion, not a nice-to-have — evaluate before you join
• Divide and conquer clearly: if someone asks you something in the CTO's domain, redirect them
• Structural cadences at Slack: async OKR reviews (video + doc per team), Monday red/yellow/green review, weekly four-person meeting (CPO + CTO + two chiefs of staff)
• When OKR reviews ballooned to 300+ person-hours per quarter, they moved everything async using Slack Clips
• Don't air disagreements behind each other's backs — be direct in the room or the relationship breaks

Designing for tomorrow's users, not today's vocal minority

• Users who complain loudest about a removed feature are often a tiny fraction of the user base
• The people using your product tomorrow outnumber the people using it today — design for them
• When you make a hard change: explain why honestly, give people time, give them choice
• Authenticity beats marketing speak; people who feel heard change their minds
• Rachel Letston (Google global PR lead) called an IC engineer directly who was upset about a policy change — he flipped completely and told the whole org

AI and what it means for product work

• People who figure out AI leverage now will be far ahead in five years — the gap is already opening
• The lines between PM, engineer, and designer will blur as AI enables prototyping across roles
• PMs who do mostly project management are most exposed; strategic, creative PMs are least exposed
• AI is not good at creative leaps — it surfaces what customers are asking for, not how to differentiate
• Real examples in use now:
  • Fed an entire Discord channel into Gemini to extract product sentiment and feature requests in minutes
  • Gong call transcripts → Glean app → structured spreadsheet of top customer-requested features (required prompt iteration to distinguish customer voice from salesperson voice)
  • LaunchCal + open Jira tickets + Slack conversations → confidence score on a feature's ship date, triggered by one prompt
  • Engineers automating incident triage by surfacing similar past incidents

Building AI-native products on non-deterministic LLMs

• Enterprise users expect determinism; chat interfaces are still unfamiliar — both require guardrails and user education
• Glean's biggest support issue early on: users asking questions the system couldn't possibly answer (e.g., "what should my top priority be next week?")
• Build guardrails that suggest what to ask, not just block what won't work
• Don't build workarounds to compensate for today's LLM limitations — those workarounds become technical debt as models improve
• Your differentiator must survive LLM improvement; if your moat is patching a weakness the model will fix, you have no moat
• The competitive frame: as LLMs get smarter, everyone's baseline rises — your advantage must come from outside the LLM itself

Staying ahead as AI moves fast

• Use the products — same rule that applied when mobile launched applies now, more urgently
• Give AI a role in the prompt before the task: "You are a PM at [company]. From that perspective, summarize…"
• Load articles before a commute; use voice mode to query and discuss during the commute
• Newsletters worth following: Ben's Bites, The Neuron
• Podcasts: No Priors (Gill and Sarah Guo), Cognitive Revolution