LLenny's Podcast
← All episodes
Shweta Shrivastava (Waymo, Amazon, Cisco)09 April 2023

Product lessons from Waymo

5Frameworks
14Insights

Frameworks in this episode

Strategy3 steps

Disrupt Yourself Before Someone Else Does

Market leaders survive by challenging and cannibalizing their own products before an upstart does it for them.

Read →
Leadership3 steps

Get Promoted by Not Optimizing for the Promotion

Focus 100% on business impact and make your ambitions known — but never engineer your work around the title.

Read →
Leadership3 steps

The Failure Interview Question

Ask senior candidates about one time they failed and what they learned — and watch how they answer.

Read →
Communication3 steps

The Rule of Seven

When an email thread hits seven replies without resolution, stop typing and get on a call.

Read →
Innovation3 steps

Write the Press Release First

Before building anything, write the launch press release as if the product shipped today.

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster11:30

Waymo Isn't Tesla: No Human Behind the Wheel by Design

Responding to Lenny's instinct to disengage Tesla's self-driving on a curvy road, Shweta clarifies that Waymo is a fundamentally different system. Waymo set out to solve fully autonomous (L4) driving with no human driver from the very beginning, so trust and predictability had to be designed in from day one — unlike a driver-assist (ADAS) system that expects the human to take over.

  • Waymo is L4 — fully autonomous with no human driver expected at the wheel
  • It is not a driver-assist / ADAS system that hands control back in complex situations
  • Because there's no fallback driver, predictability and trust had to be core to the design philosophy from the start

“we started by solving the problem of fully autonomous driving without a human driver at the wheel from the get-go right”

Shweta Shrivastava · 11:30
#waymo#tesla#autonomy-levels#trust#adas

Hot Take· 2

Hot Take28:00

Disrupt Yourself Before Someone Else Does

Drawing on the innovator's dilemma, Shweta warns that market-leading big companies get incremental in their strategy and then get disrupted by upstarts. Her conviction is that you must disrupt yourself before someone else does — it's inevitable. Her own example: launching Honeycode, Amazon's first no-code application development platform, in a space AWS had never entered.

  • Market leaders drift toward incremental strategy and get blindsided by upstarts
  • You must disrupt your own models before a competitor does — it's inevitable
  • She launched Honeycode, Amazon's first no-code app-development platform, as a self-disruption example

“you need to disrupt yourself before somebody else does right because it's gonna happen it's inevitable”

Shweta Shrivastava · 28:00
#innovation#innovators-dilemma#strategy#amazon
Hot Take33:00

The Way to Get Promoted Is to Not Want It Too Badly

Shweta's cheeky-but-serious advice on promotions: don't want it too badly. Focus on impact and doing what's right for the business rather than optimizing for the title. If you give it your 100% on the right things, the impact will be visible — and it's fine to make your ambitions known to your manager so they can hand you stretch projects.

  • Optimizing visibly for a promotion is not a positive signal and distracts you
  • Focus on real business impact; if you give 100% on the right things it becomes visible
  • Do make your ambitions and desire for challenging, high-visibility work known to your leader

“the way to get promoted is to not want it too badly right”

Shweta Shrivastava · 33:30
#career#promotion#product-management#leadership

Explainer· 3

Explainer06:00

How Waymo Gives Its Cars Human-Like Body Language

Shweta explains how Waymo trains deep-learning models on human driving data so the car's behavior feels natural rather than robotic. The system reads other road users' intent (which way a pedestrian is looking, body orientation, gestures) and also has to learn unwritten social norms that differ from one intersection to the next. Critically, Waymo discards the bad human driving data and mimics only the good behavior.

  • Deep-learning models are trained on human driving so the car doesn't feel robotic
  • The system reads pedestrian gaze, body orientation, and hand gestures to infer intent
  • Bad human driving data is discarded; only good behavior is mimicked
  • Driving is governed by social norms that vary by city and intersection, not just explicit signs

“we discard the bad human driving data uh we can mimic human driving behavior in a good way and that's why you saw the behavior…”

Shweta Shrivastava · 07:00

“we don't realize how sophisticated how interactive uh and how social driving really is”

Shweta Shrivastava · 07:30
#waymo#self-driving#deep-learning#trust#driving-behavior
Explainer16:00

The KPIs Waymo Uses to Track a Self-Driving Business

Shweta breaks Waymo's metrics into commercial/operational (trips per week, active users, funnel metrics, cost) and system-behavior (safety and progress). The core safety goal is to drive safer than humans, benchmarked against data like collisions per 100,000 miles. On the other side, they track 'stops and strandings' — because a car that's perfectly safe but never moves is its own failure.

  • Two metric categories: commercial/operational vs. system-behavior
  • Commercial: trips per week, daily/weekly active users, funnel metrics, cost to operate
  • Safety goal is to drive safer than a human benchmark (e.g. collisions per 100,000 miles)
  • 'Stops and strandings' track the opposite failure — being too cautious to make adequate progress

“the goal here is to be able to drive safer than uh than humans right”

Shweta Shrivastava · 17:30

“you can be very safe if you're not moving at all that's one problem right”

Shweta Shrivastava · 18:30
#waymo#kpis#metrics#safety#product-management
Explainer21:00

Why Waymo Doesn't Need Level 5 — and Snow Is the Real Blocker

Shweta distinguishes the autonomy levels: L2/L3 is driver-assist, L4 is fully autonomous with no human at the wheel (Waymo's focus), and L5 is driving anywhere including unstructured off-road with no map or priors. Her view is that L4 realizes the dream of autonomous driving without ever needing L5, which may stay niche. The technology is largely there — the remaining gap is capabilities like driving in snow.

  • L2/L3 = driver-assist; L4 = fully autonomous, no human at the wheel; L5 = drive anywhere with no map or priors
  • L4 can deliver the autonomous-driving dream without needing L5, which may remain niche
  • The main technical blocker isn't the core tech but edge capabilities — Waymo can't yet drive in snow

“the technology is there but it still needs improvements especially you know we were not able to drive in snow yet”

Shweta Shrivastava · 22:00
#autonomy-levels#waymo#self-driving#technology

Story· 2

Story12:30

Why a Waymo Ride Feels Boring After Five Minutes (On Purpose)

Riders consistently report that the first five minutes feel like 'wow, is this really happening,' and then it becomes completely natural. Shweta stresses this is not happenstance — the smoothness, naturalness, and constant adherence to safety are deliberately designed, backed by small trust cues like rider-support calls and an in-car monitor showing what the car sees.

  • Riders find the experience uneventful after the first five minutes — which is the goal
  • Naturalness and smoothness are engineered, not accidental
  • Trust is built from many small things: visibility into what the car sees, rider-support outreach, an in-car monitor

“it's like for the first five minutes of the ride it's wow you know is this thing really happening right but then it starts to…”

Shweta Shrivastava · 12:30

“but it's not a happenstance that it feels that way”

Shweta Shrivastava · 13:00
#waymo#trust#user-experience#product-design
Story14:30

The Surprising Lesson About Driving Below the Speed Limit on Hills

Waymo never exceeds the speed limit, which riders appreciate. But the team learned that strictly adhering to the limit isn't always the most natural feel — on San Francisco's steep inclines, human drivers subconsciously slow down going downhill. The autonomous vehicle doesn't technically need to, but riders expected it, so Waymo modified the behavior to match.

  • The car never exceeds the speed limit, which riders value
  • Human drivers subconsciously slow down on downhill gradients
  • Waymo modified its behavior to slow below the limit on SF inclines to feel more natural to riders

“it will never go up above the speed limit right it doesn't go about the speed limit it sticks or adheres to the speed limit”

Shweta Shrivastava · 14:30

“the human brain is trained to other human drivers are sort of you know subconsciously they slow down when they go downhill”

Shweta Shrivastava · 15:00
#waymo#driving-behavior#user-experience#san-francisco

Tool· 2

Tool39:00

The 'Rule of Seven' for Killing Endless Email Threads

A small process change with outsized impact: Shweta's 'rule of seven.' If an email thread hits seven messages without resolving the issue, stop typing and get on a call or into a room. Long threads that don't converge waste many people's time; at bigger companies she stretches the limit to around ten, but the principle holds.

  • If a thread reaches ~7 emails unresolved, switch to a call or huddle
  • Non-converging threads waste many people's time
  • The limit can flex to ~10 at bigger companies, but the trigger is the same

“if there have been seven emails in an email thread and you still haven't resolved issue just call the person or get in a room”

Shweta Shrivastava · 39:30
#productivity#communication#process#teams
Tool36:30

Shweta's Two Product Management Classics

In the lightning round, Shweta names the two product management books she's quoted and recommended most: Crossing the Chasm by Geoffrey Moore and The Innovator's Dilemma by Clayton Christensen.

  • Crossing the Chasm by Geoffrey Moore
  • The Innovator's Dilemma by Clayton Christensen
  • Both are the PM classics she has recommended to many people

“crossing the Chasm by Jeffrey Moore and uh Clayton christensen's uh innovators dilemma”

Shweta Shrivastava · 37:00
#books#product-management#recommendations

Takeaway· 4

Takeaway10:00

Why 'MVP' Means Something Completely Different at Waymo

The classic SaaS idea of shipping a minimum viable product and iterating in the real world still applies at Waymo, but the safety bar changes everything. Because you can't cut corners on safety, the MVP threshold itself is extremely high. PMs also have to go deeper technically and be comfortable playing a long game.

  • The MVP-and-iterate philosophy applies, but the safety bar for the MVP is extremely high
  • You can't cut corners on safety, so 'minimum' means something very different
  • PMs must go as technically deep as needed and tolerate uncertainty
  • The mission — reducing the ~1.35 million annual traffic deaths — drives self-selection into the role

“the MVP bar itself for safety is extremely high for us right”

Shweta Shrivastava · 10:30
#product-management#mvp#safety#waymo
Takeaway23:00

How to Keep Leaders Bought Into a Multi-Year Investment

Asked how Waymo keeps Alphabet and investors committed to a years-long bet, Shweta says the name of the game is showing meaningful progress — not just technology milestones but commercial deployments. Results have to speak for themselves. The deeper lesson: focus on building the right business and creating value for customers rather than chasing short-term brownie points, which investors see through anyway.

  • Show meaningful progress in both technology and commercial deployments
  • Let results speak for themselves; that's where the rubber meets the road
  • Don't do unnatural things for short-term investor points — they see through it
  • Lenny's takeaway: it's hard to cut something that keeps showing success

“the name of the game here is to uh is is to show progress right show meaningful progress and meaningful progress not just in terms…”

Shweta Shrivastava · 23:30
#fundraising#product-strategy#leadership#long-term-bets
Takeaway27:00

The Underrated Discipline of Knowing What You're NOT Building

One of Shweta's biggest cross-company lessons is that it's just as important to know what you're not building as what you are. It's easy to get swayed by different customers pulling you in different directions, but a product that tries to be all things to all people usually goes nowhere. Focus and crisp prioritization are essential in both big companies and startups.

  • Knowing what you won't build is as important as knowing what you will
  • Different customers will pull the roadmap in conflicting directions
  • A product that tries to please everyone usually goes nowhere

“it's also very important to know what you're not building right”

Shweta Shrivastava · 27:00

“a product that tries to be all thanks to all people you know usually doesn't end up going anywhere”

Shweta Shrivastava · 27:30
#product-management#prioritization#focus#strategy
Takeaway29:30

The Most Underrated PM Skill: Listening, Empathy, and Challenging Your Own Assumptions

Shweta's pick for the most important early-career PM skills is listening and empathy — easier said than done, requiring a beginner's mindset where you absorb before jumping in with ideas. Her key tell for whether you're really listening: are you challenging your own assumptions? If there's no conflict or contention in the conversation, something is missing.

  • Listening and empathy are core, not soft, skills — but hard to actually do
  • Come in with a beginner's mindset; absorb and learn before pitching ideas
  • Actively challenge your own assumptions as a test of whether you're truly listening
  • If there's no conflict or contention, you're probably missing something

“the listening and uh and empathy are are the top ones”

Shweta Shrivastava · 29:30

“if there's no conflict if there's no contention then something is missing”

Shweta Shrivastava · 33:00
#product-management#career#listening#empathy