YT YouTube2Text

YouTube2Text → Transcripts → Why Netflix is betting on systems thinkers—not specialists—in the AI era | Elizabeth Stone (CPTO)

Why Netflix is betting on systems thinkers—not specialists—in the AI era | Elizabeth Stone (CPTO) — Transcript

by Lenny's Podcast · 13,484 words · 2,115 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Everyone can be everything now. PMs can
0:01ship code, designers can write PRDs,
0:03engineers can product, and there's this
0:05confusion and frustration of what is my
0:07job anymore.
0:08>> Anytime a new technology comes along,
0:11you go through a storming phase before
0:14you go through the forming phase of
0:16things. We are in the middle of that
0:17right now. [music] I don't think that
0:19means we should put AI back into the box
0:22and say let's not use it.
0:23>> If we all become builders, will we still
0:24need separate functions?
0:26>> I still see a craft excellence that's
0:28really important that [music] I don't
0:29think is going away anytime soon. I
0:31still find great engineering to be
0:34scarce, great data science to be scarce,
0:36great creativity to be scarce.
0:38>> If you look at the early culture deck of
0:40Netflix, high agency, autonomy, paying
0:43top of market, this is what I hear
0:44constantly now from how the top AI labs
0:47operate.
0:47>> Netflix's culture has always been
0:49excellence as an operating system. It's
0:51a resistance [music] to do the thing
0:52that a lot of bigger companies would do
0:54and to feel comfortable in that
0:57discomfort very often.
0:58>> What are the ingredients to make this
1:00happen?
1:00>> Talent density is the non-negotiable,
1:03being very comfortable with risk-taking
1:04in cases where things are not going
1:06well, not assume that process is going
1:08to fix it.
1:09>> What have you added to the career
1:11ladders within this AI world?
1:13>> more systems thinkers, people who can
1:16look across all the business domains and
1:19abstract that [music] to here's the
1:20building blocks we're going to need.
1:22>> How do people learn this?
1:23>> Small trick, each problem you're trying
1:25to solve, step out one [music] click to
1:28the what am I assuming is true about the
1:31broader space.
1:34>> Today my guest is Elizabeth Stone,
1:36product and technology officer at
1:38Netflix. This is Elizabeth's second
1:40visit to the podcast. Her first visit,
1:42when she was just a CTO, was for the
1:44longest time one of the most popular
1:46episodes of this podcast. You'll soon
1:48see why this is such a killer
1:50conversation because when we chatted two
1:52and a half years ago, AI was only
1:54starting to emerge. [music] And as a
1:56long time head of engineering and
1:57product and data science, Elizabeth has
1:59such a unique perspective on where
2:01things [music] are heading and what's
2:02worth paying attention to. Prior to
2:04Netflix, Elizabeth was VP of Science at
2:06Lyft, Chief Operating Officer at Nuna,
2:08[music] an economist at The Analysis
2:10Group, and a trader at Merrill Lynch.
2:12Before we get into it, don't forget to
2:13check out Lenny's Product Pass dot com
2:15for an entire year free of the hottest
2:18and best crafted AI products in the
2:20world available exclusively to Lenny's
2:22newsletter subscribers. With that, I
2:24bring you Elizabeth Stone.
2:29Elizabeth, thank you so much for being
2:31here and welcome back to the podcast.
2:33>> Thank you. I'm honored to be here. Once
2:35and now twice.
2:36>> That's right. That's a rare a rare treat
2:38for me. I don't know if you know this,
2:40but your first visit to the podcast,
2:43your episode ended up being my second
2:45most popular episode. You're right
2:47behind Brian Chesky for the longest
2:49time.
2:50>> Well, I I I'm pleasantly surprised and
2:53also mildly competitive of how
2:56[clears throat] do I get to the first
2:57spot? But I'll set that aside for now.
2:59>> That's This is our This is our shot.
3:01>> Bri- Brian's amazing, so I'll let that
3:03one go.
3:04>> Yeah, he is uh and then there's just
3:06like all these fancy AI people that are
3:07just coming, you know, coming in hot.
3:09>> [laughter]
3:10>> Um so, it's been 2 and 1/2 years at this
3:12point. A lot's changed.
3:14Uh obviously AI, something AI is
3:17allowing uh people to do is everyone can
3:19kind of be everything now. This idea of
3:22PMs can ship code, designers can write
3:24PRDs, and engineers can product, and
3:26everyone's everything. There's a bunch
3:28of
3:29elements to this conversation. One is
3:31that I've heard from people that there's
3:33also this kind of confusion and
3:35frustration of like what is my job
3:37anymore? Like what am I responsible for
3:40as a PM, as a designer? Is that
3:42something you've experienced?
3:43>> I hear it within Netflix, for sure.
3:47I think anytime a new technology comes
3:50along, especially one that's as
3:52transformative as GenAI,
3:55you go through a storming phase before
3:58you go through the forming phase of
3:59things. And I think we are in the middle
4:01of that right now.
4:03I don't think that means we should put
4:06AI back into the box and say let's not
4:08use it cuz this is kind of this is
4:10complicating all of our preconceived
4:12notions about our roles,
4:14but I do think it means we have to be
4:15much more thoughtful about how do we get
4:17the benefits while reducing the costs.
4:20I think it's a great thing that people
4:22are experimenting with how can I develop
4:25an idea faster, prototype an idea, put
4:28together an initial set of code that
4:30would allow us to test it.
4:32Do I believe that means anyone should be
4:35shipping code to production?
4:37That everyone should actually be doing
4:39everything? Probably not. But I think
4:42that it's good for people to be
4:43exploring what's possible. And then,
4:45like I mentioned earlier, the benefit of
4:47having product and tech teams together
4:50is that if the business problem is
4:52clear, I think it's okay and it's
4:54healthy for there to be some fluidity in
4:56the roles that people play because
4:58instead of having to wait for the
5:00engineering team to be ready to be able
5:02to prototype something, product and
5:04design can move faster on it. But they
5:06should still work with their engineering
5:08partner to think through how should we
5:09productize this? How do we scale it?
5:11What are the guardrails for it? So, I
5:13don't think it makes the functional
5:15expertise obsolete. I think it means
5:18that teams have to be more comfortable
5:19with maybe this helps us move faster in
5:22a certain direction. From an
5:23organizational perspective, things I
5:26think about to make this
5:29more coherent or less frustrating
5:31are some of the things that have to be
5:33in place for us to get the benefits
5:36rather than the costs. So, that includes
5:38clarity on source of truth data,
5:40guardrails on shipping code to
5:42production or testing before we make
5:44large changes,
5:46thinking about opportunities where we
5:49can trust the output of AI versus we
5:51should have a process or review that
5:53helps us check that we're getting high
5:55quality outcomes.
5:56And the importance of reiterating that
5:59humans are still responsible
6:01for what happens. So, it can be that an
6:03agent wrote the code or I helped to do
6:05an analysis when that's not really my
6:07background, but it doesn't make it
6:09doesn't
6:10make people not have the responsibility
6:13that comes with what they've created.
6:15So, I think the investing in some of
6:17those core infrastructure and practices
6:19and reiterating the accountability and
6:20responsibility for the outcomes helps to
6:23balance some of like what's possible
6:25with what we should actually be doing.
6:27>> This episode is brought to you by our
6:29season's presenting sponsor WorkOS. What
6:32do OpenAI and Vercel,
6:35Replit, Sierra, Clay, and hundreds of
6:38other winning companies all have in
6:39common? They are all powered by WorkOS.
6:42If you're building a product for the
6:43enterprise, you've felt the pain of
6:45integrating single sign-on, SCIM, RBAC,
6:48audit logs, and other features [music]
6:50required by large companies. WorkOS
6:52turns those deal blockers into drop-in
6:54APIs with a modern developer platform
6:57built specifically for B2B SaaS.
6:59>> [music]
6:59>> Literally every startup that I'm an
7:01investor in that starts to expand
7:03upmarket ends up working with WorkOS.
7:05And that's because they are the best.
7:07Whether you are seed-stage startup
7:09trying to land your first enterprise
7:10customer or unicorn expanding globally,
7:13WorkOS is the fastest path to becoming
7:15enterprise ready and unlocking growth.
7:17It's essentially Stripe for enterprise
7:19features. Visit workos.com to get
7:22started or just hit up their Slack where
7:24they have actual engineers waiting to
7:25answer your questions. WorkOS allows you
7:28to build faster with delightful APIs,
7:30comprehensive docs, and a smooth
7:32developer experience. Go to workos.com
7:34to make your app enterprise ready today.
7:37>> What's really awesome about having you
7:39back on the podcast is
7:41we chatted like before AI was a massive
7:43transformation in the world. So, it's a
7:45really cool arc that we can explore
7:47here. This the shift that we've all gone
7:49through.
7:50>> Mhm.
7:51>> Coming back to the roles of the product
7:54and inch team, I'm curious
7:56how much these roles have changed in the
7:58last two and a half years. If you think
7:59about product engineering,
8:01uh design, data science, user research,
8:04which roles have
8:06changed most? Which roles have changed
8:08least? Like, what's most different in
8:10the last two since two and a half years
8:12ago?
8:12>> So, you've mentioned some of the things,
8:14so I'll I'll reiterate them and then
8:16maybe build. So, I have found that PMs,
8:20designers,
8:22data scientists are able to get farther
8:26in the product development life cycle
8:29before engineering really needs to be
8:32front of the line in unlocking things
8:34than was true a couple years ago.
8:37I say that with some caution because,
8:39like we were talking about, I don't
8:41think it's great to all of a sudden have
8:43thousands of prototypes if they're not
8:45aimed at this is an important problem to
8:48solve for the business
8:49and the engineering partners are aware
8:51that we're solving that problem and that
8:53designers and product managers are going
8:54to take the lead in starting to shape
8:56the idea, but it's not working in a
8:58vacuum and it's not throwing a bunch of
9:01spaghetti at the wall to see what
9:02sticks.
9:03But when it's the right problem,
9:04approached in a thoughtful way with some
9:06alignment on that, I've seen product
9:09design data science move faster in the
9:11direction of let's get to something
9:13that's testable on this hypothesis.
9:16So, that's prototyping, that's writing
9:17code. The other thing I've seen as being
9:20very valuable is we have a lot of
9:21information
9:23running around in the virtual walls of
9:24Netflix. We have experiments we've run
9:27over decades. We have insights from
9:29consumers. We have input from
9:31stakeholders across the business. And
9:34that was a problem that really presented
9:36a challenge of like, how do we get the
9:38most out of that long history of
9:41knowledge and learnings to say, let's
9:43apply that to the problem we've got now
9:45to move faster in this is a promising
9:48path or this is something that we've
9:50learned something about and we could
9:51leverage here.
9:52And AI is very powerful at distilling
9:55information,
9:57looking across a broad set of things,
9:59doing an analysis around it, getting to
10:01the core of here's some insights to
10:03start with. I I would hesitate to rely
10:05on that exclusively, but I think it's a
10:07head start. And I find even in my own
10:10work day-to-day, instead of
10:12sending an email that disrupts someone
10:14of like, remind me what research did we
10:16do in what year and what was the
10:18question and what was the test we ran? I
10:20can find that almost instantly. Then I
10:22can form my own, here's what I find
10:24interesting about this and I've now
10:26skipped a couple steps towards is there
10:28something actionable here? So that's
10:30data analysis, it's modeling, it's
10:32distillation of information and I'm
10:34seeing more people do that to your
10:36original question. So instead of that
10:38needing to be
10:39only the experts who were here for 20
10:42years and saw every experiment or know
10:43where to find it, we're now able to do
10:46that faster within product and tech
10:48across all functions and a big unlock
10:50for us is our business stakeholders
10:52sitting in finance and content and
10:54advertising can do that as well and then
10:57bring back an initial hypothesis where
10:59they want to work more deeply with the
11:00data scientists and engineer and so on.
11:02So there's something there about the the
11:05hypothesis generation, prototyping,
11:08thinking deeply about problems that
11:10feels like it's accelerating and that
11:12functions are able to do that in a more
11:14fluid way.
11:16But I still see comparative strengths.
11:18So data scientists are still going to be
11:20experts at can we trust this data? Are
11:23we interpreting it the right way? What's
11:25the data versus judgment that we should
11:27be applying here? A product manager is
11:29still going to be exceptional at saying,
11:31have we really framed the what of this?
11:33Like the problem we're solving in the
11:35right way? An engineer still has a craft
11:38around the how. How does this scale?
11:40What does high quality look like? What
11:42problems is this going to create for us
11:44based on how we build and deploy
11:46something? So I still see the nuggets of
11:48that comparative advantage. It's just
11:50that we're able to move more fluidly in
11:52a lot of steps that normally we would
11:54have blockers on.
11:55>> There's so much interesting stuff here.
11:57One is this last point you made is
11:58something I've been thinking about. If
12:00we all become builders, will we still
12:02need separate functions? There's this
12:04like member of technical staff trend
12:05that is happening in the past where it's
12:07like, all right, we don't have a title,
12:09you could be anything.
12:10You don't have to be in a bucket. What
12:11you're saying here is you believe we
12:13will continue to have specialties,
12:15product person, engineer, data science,
12:17designer.
12:19While they do more of other functions,
12:20there's still a lot of value in Tell me
12:23if I'm hearing you correct in having the
12:24specific discipline and skill and
12:26background.
12:26>> I still see a craft excellence that's
12:28really important in the disciplines that
12:31I don't think is going away anytime
12:33soon. Even if there's fluidity or
12:35blurring of the work across the
12:37functional lines. It goes back to what I
12:39mentioned earlier of you still have
12:41humans who have to make sure that what
12:43we're doing makes sense. We're solving
12:45the right problems in a way that is best
12:47for Netflix members or business
12:49stakeholders.
12:51And that if I talk to an engineer, a
12:54data scientist, a designer,
12:57yes, they speak more languages now than
12:59they used to because they have the
13:00benefit of these AI tools.
13:03But there's still something that is not
13:05replaceable
13:07when I think about the craft and how
13:09they think about what good looks like.
13:11And that feels true across all levels
13:13and you know, I still find great
13:15engineering to be scarce. Great data
13:18science to be scarce. Great creativity
13:20to be scarce. So I
13:22yes, some things are easier, but that
13:25hasn't dissolved in my mind.
13:27>> Are there functions that you are finding
13:30you are hiring more of? Like the pie
13:33chart pie expanding
13:35say for engineering or PM or design or
13:37something and then functions you're need
13:39less of with AI tool and LLMs rising.
13:43>> Not sure that it matches exactly to
13:45functions, but I can tell you
13:48what we're having we're seeing more of,
13:50we need more of.
13:53We need more systems thinkers
13:55in a world with AI.
13:57That looks a little bit different across
13:59functions, but I could play out a couple
14:01examples. So,
14:03in our core infrastructure team at
14:05Netflix in central engineering,
14:09a lot of what made Netflix successful
14:11over time was that
14:14local teams with specific business
14:16problems could move fast to deliver.
14:20They very often were not feeling like
14:23they needed to be on a central paved
14:24path. They built the stack that they
14:26needed to solve the problem and have the
14:28impact.
14:29In a world of AI with agents operating
14:32across multiple systems,
14:35wanting source of truth data, the
14:36importance of having preferred paved
14:38paths that
14:40get the most of the benefits and produce
14:42some guardrails so we can make sure
14:43we're doing good work,
14:45common infrastructure, common paved
14:47paths, solving problems once with a core
14:50set of capabilities becomes more
14:52important.
14:53So, we are hiring more people who can
14:55look across all the business domains and
14:58abstract that to here's the building
15:00blocks we're going to need in a world
15:02with AI. So, that's one of the lenses,
15:04but also just with a lens of what got
15:07Netflix here doesn't get Netflix there.
15:09And we're going to have to have a
15:11stronger set of infrastructure to move
15:13quickly in this future.
15:14So, that means that engineering profiles
15:16are more distributed systems, more
15:18infrastructure, more of that system
15:20thinking mindset than a a local business
15:22expertise. Though, of course, we still
15:24have people who are deep in
15:26personalization and advertising and
15:28content delivery. So, it's more
15:30something additive for us to have that
15:32core infrastructure and systems
15:34thinking.
15:35If I take another example, like design,
15:39it's extremely important that our
15:41experience design team is developing
15:44templates and again systems thinking for
15:47what does great user design look like at
15:49Netflix so that they can enable lots of
15:52people, including those who are not
15:54designers by training, to develop
15:56products that are coherent, that fit
15:59into the end-end member experience. I
16:01get really nervous about having
16:03different design languages or different
16:04types of user interactions and shipping
16:07Frankensteins, basically. So, designers
16:10need to then be the people we're hiring
16:13again for design systems thinking. How
16:15do we think about templates and
16:17expression of the brand and what a good
16:19user experience looks like and what is
16:21Netflix and like the Netflix
16:22differentiated special sauce. So,
16:24there's more people on our design team
16:26that have to think that way now than
16:29could I help to design a specific
16:31feature for a specific product. So,
16:33there's this stepping back to look at
16:35the big picture that I think is
16:36happening in every single function and
16:38that requires
16:40some, yeah,
16:42reorientation of skills among the
16:43existing team and also hiring people
16:46who've got that that type of expertise.
16:49And across all of it, it's a mindset
16:51shift. So, we are not hiring people
16:55who are not excited to explore, try new
17:00things, understand lots is changing and
17:02feel comfortable with that ambiguity,
17:05be comfortable that there's a blurring
17:06of how we work and how we partner. It
17:09that's true for people who are already
17:11at Netflix and people who we are adding
17:13to the team that that curiosity
17:15innovation mindset has not
17:18it's not been more important, at least
17:19in the time that I've been working in
17:21this field.
17:22>> On the systems thinking piece, is the
17:24reason this is becoming more important
17:26that it is people are moving so fast
17:28that you need to invest in platforms and
17:31frameworks and and design language and
17:33basically
17:34teach people to fish so they can not be
17:36blocked or is there is there other
17:38reasons?
17:38>> I think it's probably velocity. So
17:40platforms do have a benefit of leverage.
17:43So in general, that that's an
17:45opportunity with or without AI for a
17:47platform to get most teams 80% of the
17:49way there.
17:51And then they don't have to reinvent
17:52those building blocks.
17:54We have more bets that we're making
17:57across the business, more things we're
17:58trying to build. So platform mindsets
18:00are good and it's something that is
18:02relatively more recent for Netflix to
18:04think about that being a real critical
18:07enabler.
18:08There is also the sense of a scaffolding
18:12in a world of AI. So not just the higher
18:14velocity, but you have more people doing
18:17more types of work that are different or
18:19new like we were talking about. And
18:21there's risk that comes with how do you
18:23think about access and identity in that
18:26situation? How do you think about
18:27security in that situation? How do you
18:29think about how shipping high quality
18:31code and design and user experiences?
18:34And so I I don't think it scales well to
18:37have each person who's building
18:38something have to go figure out. Could
18:40you remind me what good looks like here
18:42and what are the bumpers or guardrails I
18:44should keep in mind? I think we need to
18:46encode that in our paved paths and our
18:49ways of working. And for a a data
18:51science or analytical field to encode
18:54here's the source of truth data, here's
18:55how to interpret it, here's how to
18:57access it, here's what to do with it or
18:58not to do with it and to be careful with
19:01certain types of data.
19:02I don't an organization that has
19:04thousands of people can no longer rely
19:06on tribal knowledge or I'm going to find
19:08the one person who knows this. So this
19:10was a challenge that was there before
19:12AI. It's probably a more urgent
19:14challenge with AI and I like the idea of
19:17using AI or any new tech to motivate
19:20like we knew this is work we needed to
19:22do. No time like the present to invest
19:25in that more heavily across the team.
19:27>> I wonder if another reason for this
19:29becoming more valuable is because agents
19:31are now doing a lot of work and giving
19:33them the context, giving them the
19:34scaffolding, giving them the design
19:35language just speeds all that up.
19:38>> Yeah, and one of the visions we have at
19:40Netflix is
19:41we will have so many
19:43agents that are contributing to doing
19:46work that you need to be able to reason
19:48and rationalize throughout that. You
19:50know, the humans are the ones guiding
19:52what's the problem we need to solve. Do
19:54I feel like what we're producing is
19:56impactful and high-quality output?
19:59But the work will be done by both humans
20:01and agents.
20:02And that creates velocity and benefits
20:05and it creates risks. And I think that's
20:07important from especially from an
20:09engineering perspective that we figure
20:11out how to manage that in a way that
20:13lets people move quickly but doesn't
20:16create undue downside or risks for the
20:19company.
20:20>> This connects so directly with
20:23Jenny Wen was on the podcast. She was
20:24head of design for Cloud Code and
20:26Co-work and had this whole design
20:28process is dead kind of thesis and the
20:30pitch there is just there's no time for
20:31design, the design process. And instead
20:34as a designer, you're just kind of
20:35steering people and pointing them in the
20:37direction
20:38and adjusting and also thinking big
20:40picture is when you have the time.
20:43And it feels like that's kind of what
20:44you're describing here is like create
20:45the platform for people to move fast and
20:47then there's no time for like design
20:49process of a specific new feature.
20:51>> I have mixed feelings about that because
20:53I
20:53we do want to enable with infrastructure
20:57and systems thinking more people to do
21:00great work with strong design as part of
21:03it.
21:04Why not take that opportunity that the
21:06new tech provides.
21:08But for our most important priorities,
21:12design is critical
21:14to solve things in the right way. So, we
21:17do still make time for important design
21:19work. We It can move faster. The
21:21designers themselves have more tools in
21:23their toolkit, so they can do incredible
21:26work at a faster velocity, show more
21:28options, learn, iterate, test more
21:31quickly. But I think it would be a
21:33mistake to say
21:35design and deep design expertise and
21:37thinking gets squeezed out just because
21:39we can write code faster. We can do data
21:41analysis faster. That feels like, at
21:44least for a large-scale consumer product
21:46like Netflix, I feel like we would lose
21:48one of the things that makes Netflix
21:50great, which is the product, technology,
21:53and design makes a lot of complexity
21:55invisible, and makes for a seamless
21:57customer experience. That That's a
21:59design mindset that has to be core to
22:02it. So, if the work itself might look
22:03different, but I don't think we lose the
22:05mindset.
22:06>> That's an awesome counterpoint.
22:08So, what I'm hearing is kind of trending
22:10up skills, attributes you look for,
22:12systems thinking, and this kind of
22:15mindset of being comfortable and excited
22:17about change and what's coming and not
22:19being stuck in your own ways.
22:20What are you finding is trending down?
22:23What are you less looking for that used
22:26to value more highly?
22:28>> The days of very narrow, deep
22:31specialization
22:33feel more limited to me.
22:35I can come up with examples where we
22:37still need it because there's an
22:40industry or technology expertise where
22:42there's only a few people in the world
22:44who really know how things work. We have
22:46examples of that on the team for
22:48encoding or how our playback systems
22:51work and things that have been
22:53incredibly innovative and novel for
22:55Netflix. I I still believe we need
22:57specialized practitioners in those
22:59spaces.
23:01But as a general rule, uh compared to 5
23:05or 10 years ago, I I would believe we
23:07have fewer specialists and more people
23:09who are generalists or adaptable in
23:12multiple directions. And that could be
23:14adaptable across functional expertise.
23:17It could be adaptable across flavors of
23:20engineering. So, can I navigate both
23:22back end and front end systems? Can I
23:25hook into infrastructure with a lot of
23:27expertise? I think
23:29the the mindset now needs to be I can
23:31learn that quickly, and that goes back
23:33to the systems thinking. So, I think
23:35specialists can learn to have a broader
23:37array of tools more easily than was true
23:40in the past. So, it we need fewer of
23:43them perhaps because talent's able to
23:45grow in that direction. And there's
23:48something about sticking to a narrow
23:52specialty that maybe triggers for me a
23:55concern about what about the mindset of
23:58growing in different directions and
23:59exploring boring, and I don't want to be
24:02too narrow even in my own assessment of
24:03that, but I it's important that people
24:05who are specialists still have that
24:07sense of I want to try a new way of
24:09solving these problems versus the way we
24:11have in the past.
24:12>> And when you say specialist, are you
24:13thinking like front end, I'm a front end
24:15engineer versus a back end, or are there
24:16other
24:17>> Yeah, or it could be a domain set of
24:20knowledge of Yeah, I'm a deep
24:22>> expert.
24:22>> I'm a payments expert. I'm an
24:25ads marketplace design expert. I'm an an
24:29expert in this very specific tooling
24:31that studio productions use.
24:34>> Mhm.
24:34>> So, there
24:37specialist in subject matter expertise
24:39is an advantage provided that person is
24:43willing to grow and extend into is this
24:46really still the right tool or the right
24:48way to think about the problem? So, I
24:50think it's the layers of the stack from
24:52an engineering perspective that there's
24:54less specialty.
24:55And then
24:57tools that are unlikely to be static or
24:59like to have a lot of inertia around
25:01them. I would think like we would want
25:03people who are able to innovate and
25:05imagine like what's the future version
25:06of this? And so we want more talent like
25:08that.
25:09>> Awesome. So coming back to the systems
25:11thinking piece, people hearing this are
25:13like, okay, I got to work on my systems
25:14thinking
25:15skill set. How do people develop the
25:17skill? Other Is it just do it for a long
25:19time? Work at a lot of complex
25:21projects? Like I think of this book that
25:23everyone always references with the
25:25slinky on the front, Thinking in
25:26Systems.
25:28>> [laughter]
25:28>> Yeah, how do people learn this?
25:31>> Small trick.
25:33Each
25:35problem you're trying to solve, step out
25:38one click.
25:40Do the like, what am I assuming is true
25:43about the broader space in solving this
25:45problem?
25:46So I was given a task to build some new
25:50feature for the Netflix member
25:52experience.
25:53Let me take one beat and think about
25:56what is the bigger consumer problem
25:58we're trying to solve here?
26:00What's the type of content that this
26:01feature is going to be able to support?
26:05Do I think that the way I was planning
26:07to build this is going to make sense in
26:09a way that scales across multiple
26:11content types? Or it could be something
26:13that's a capability that then is
26:15contributed to a platform set of
26:18offerings for multiple areas.
26:20Is the consumer problem that I'm solving
26:22with this feature
26:24going to be one of the most important
26:27consumer problems that Netflix is going
26:28to need to solve as we have an expanding
26:30world of entertainment and we want to
26:32make it more personalized and immersive.
26:34Those are all questions that like you
26:36don't have to boil the whole ocean. You
26:38don't have to solve for Netflix's
26:39overall strategy and who are we relative
26:41to competition.
26:43But you take the thing you're
26:44responsible for and you just do one zoom
26:47out of the problem you're solving and
26:49question that.
26:51I wouldn't spend too long in the
26:52questioning state because then you're
26:54stuck. Then you're not making forward
26:55progress, but I think that helps people
26:58to think in terms of systems and
27:01question that are we solving the right
27:02problem in the right way that matters
27:04for the end consumer.
27:06>> Another way as you describe it, another
27:08way I'm thinking about it is like think
27:10if you were your manager
27:12how would they what's their broader
27:13perspective across not just your one
27:15team and problem and KPI, but the larger
27:17picture?
27:18>> I've got advice over years that is
27:20similar to that which is
27:22are there ways that I can do my job that
27:25helps
27:27my manager do their job.
27:30And so if I thought about all the things
27:31I'm directly responsible for, but I
27:33thought about it from the perspective of
27:35my manager. So not just product and
27:37tech, but finance and content and other
27:39parts of the business, I would naturally
27:42zoom out and think about how all these
27:44component pieces need to come together
27:46and how the whole could be greater than
27:47the sum of the parts. I think that's
27:49useful thinking. And for engineers to
27:51think about how do I leave a better
27:53version of these systems? How do I think
27:55about the thing that's going to be high
27:56quality and scale for others? There's
27:59both a how do I help my manager and
28:01there's how do I help my colleagues,
28:02which is a core part of some of our
28:03engineering principles of
28:05do the thing that is right for the
28:07broader organization instead of just
28:09what's right for you locally. That's
28:10systems thinking as well. So it's not
28:12just seniority, but it's breadth of the
28:15way I solve this problem and I build
28:16this, is it going to be useful to my
28:18colleagues and am I going to leave a
28:19stronger version of things for the
28:21future set of innovations that we want
28:23to make?
28:24>> That is an awesome tactical advice. Uh
28:27making your manager's life easier is
28:28always a good a good tactic.
28:30>> Career-wise, several reasons. Yeah.
28:32>> [laughter]
28:34>> Following the thread a little bit
28:35I know you all added career ladders and
28:38levels recently. It was like a new thing
28:39you guys used to not have these things.
28:41So kind of all on that thread
28:43what have you added to the career
28:46ladders within this AI world. If
28:48anything that you find you want people
28:50to lean into more, you're looking to
28:52more or or not. Like, did you not change
28:55your career ladders and performance
28:56you know, criteria?
28:58>> So, the way we've approached this so far
29:00is instead of trying to articulate
29:04at each level
29:06exactly how AI changes those
29:08expectations, to instead put an overlay
29:11across all of the talent at Netflix,
29:13people on the team, and those who are
29:15hiring to talk about an aspiration for
29:18AI fluency.
29:20And what that looks like is going to
29:21vary by function. It's going to vary
29:23based on where you are in your career.
29:25That could be what level you're in or
29:26what type of role or persona work you're
29:28doing.
29:30But the aspiration for AI fluency, which
29:32is a tough thing to define.
29:35So, does it mean that I have an
29:36experimentation mindset? Does it mean
29:39that I know where AI is useful and not
29:41useful? Does it mean that I've actually
29:43built things using AI? I feel like the
29:46the way that has shown up in career
29:48ladders and how we talk about it evolves
29:50almost by the quarter, if not month or
29:53day, because the tech itself is
29:55advancing so much. So, the most useful
29:58thing is not to make it level specific
30:00or role specific, but to encourage
30:02everyone towards the expectation on AI
30:05fluency, which doesn't mean use it as a
30:07tech for the sake of tech. It's tech
30:08where it's useful, to have good judgment
30:10about that, and to have the mindset to
30:12be open-minded to explore and try new
30:14things. That's the non-negotiable for
30:17all roles, and that's true at the senior
30:19most levels of of Netflix, where we talk
30:21about we too need to have deep fluency
30:23in AI, even if we're not writing code as
30:26part of our day jobs. So, that's that's
30:28changed, and then that's showing up in
30:30our hiring practices as well. Getting
30:32comfortable within interviews exploring
30:35how are people thinking about AI or
30:37technology? What are they using in their
30:39day-to-day or their current job? How
30:41comfortable are they with change and
30:43exploration? And even for things like
30:45coding interviews, allowing candidates,
30:47of course, to use AI tools because
30:49that's going to be part of what the work
30:51requires now. So, those have been shifts
30:53that we've made, but I I doubt it's a
30:55shift that's done versus we're right in
30:57the middle of it.
30:59>> And she's going to keep following this
31:00thread. Obviously, AI is transformative
31:03for coding.
31:04It's a big unlock for prototyping.
31:08Are there other
31:09use cases of AI at Netflix that have
31:11been really impactful that people may
31:13not
31:14think about or not realize?
31:16>> So, there's two that come to mind. So,
31:17the first is
31:19data analysis, distillation of
31:21information, modeling, which is, you
31:24know, get using the tools to get our
31:25arms around all the insights we have,
31:27similar to what I mentioned before. What
31:29experiments have we run? What are the
31:31metrics that I should be looking at for
31:33a certain problem? What's the consumer
31:34research that we've done?
31:36And that is much higher velocity and
31:40much higher quality,
31:42contingent on
31:44you check that the results are valid,
31:46you work with your local data scientist
31:48and am I using the source of truth data
31:50on this?
31:51But, that's been a great one and that's
31:52one personally that I would say I most
31:55use some of these tools for. So, that
31:58goes beyond prototyping and coding to
32:00general analytical thinking and
32:02translating data to action and insight.
32:05The other one is on the
32:08content production, creation part of the
32:11business, which has lots of
32:13applications. This was true before
32:14GenAI. So, ML and AI were deeply used in
32:17a lot of the production tools. We've
32:20used them to think about how to create
32:21promotional assets at scale, how to
32:23localize in subtitles and dubs. So,
32:27GenAI is a big step function in where
32:29the impact can be in creative ideation.
32:33We call those things like
32:34pre-visualization or basically bringing
32:36a creator's vision to life before you
32:38even get into the you bring people to a
32:41set and start to actually go through the
32:43production itself.
32:44There's lots of use cases in
32:46post-production.
32:47So we recently acquired a company Inner
32:49Positive that was started by Ben Affleck
32:52that built a set of models and
32:54capabilities that allow you after you've
32:56shot something to relight, reframe,
32:59reshoot, change dialogue in ways that
33:02are very impactful to get higher quality
33:05content are still led by the filmmaker
33:07creator saying, you know what? I would
33:08like to try something else to bring this
33:10vision to life. But that impact is
33:12extremely promising and we're seeing
33:14lots of productions
33:16leverage different tools, some of them
33:17built in-house, some of them that we
33:19enable through other vendors for those
33:21content creation use cases. And then as
33:23we think about how content comes to the
33:25product, I mentioned localization,
33:27subtitles and dubs, but also how we
33:30create high-quality trailers, images,
33:34artwork at scale that then we can use to
33:37help make sure that titles find their
33:38audiences around the world. Those all
33:41are huge levers when we think about the
33:43AI impact. So that that again goes well
33:46beyond prototyping or coding to some of
33:48the creative use cases and you can
33:50imagine that just like they work for
33:52studio productions for film and TV, they
33:54work for advertising, they work for
33:56marketing, off-service campaigns and so
33:59those are all areas that we're
34:00exploring.
34:01>> This episode is brought to you by
34:02Mercury, radically different banking
34:04loved by over 300,000
34:07entrepreneurs. And now with Command.
34:09I've been a customer of Mercury's for
34:10over 6 years. I have never once thought
34:13about leaving. Mercury is basically what
34:15happens when banking is built by product
34:17people, not by bankers. They make it so
34:20easy, dare I say fun, to send invoices,
34:24move money around, set up virtual cards
34:26for folks on my team. Does your bank
34:28have an API, a terminal native CLI, or
34:31an AI-ready MCB server? I don't think
34:34so. And just recently they launched
34:36command, a conversational interface
34:38built directly into Mercury, which acts
34:41as your financial operator. I've been
34:43using command to transfer money around,
34:45to figure out what categories I've been
34:46spending the most money in, analyze my
34:48cash flows. And [music] just today I
34:50used it to find out how much I've made
34:52from a specific sponsor over the past
34:54year. I just asked, "How much have I
34:56made from X over the past year?" 10
34:58seconds later I have an answer. It is so
35:00freaking cool. Visit mercury.com to
35:03learn more and apply online in minutes.
35:05Mercury is a fintech company, not an
35:07FDIC insured bank. Banking services
35:09provided through Choice Financial Group
35:11and Column NA, members FDIC.
35:13>> You mentioned how Netflix has been very
35:17early to AI and ML for a long time. Uh
35:20younger people may not remember this,
35:21but y'all had this contest to optimize
35:25things.
35:26Yeah. Yeah, the Netflix prize. Like like
35:28just showed an example of how early you
35:29were to AI and ML. People There was I
35:32think it was a million-dollar prize to
35:34optimize the Netflix ranking algorithm a
35:36little bit. Like whoever could optimize
35:37it the most. And I think the winner
35:39optimized it by a few percentage points,
35:41something like that. And it was like the
35:42a huge deal. All these super smart
35:44people got around around the world. Uh
35:47and it happened a few times, right?
35:49>> I mean, you said it on my behalf. Um
35:51often when there's questions about how
35:54is Netflix thinking about AI, it's great
35:56to remind people of exactly that point,
35:58that this is not new to us, that
36:01especially for personalization, it's
36:03been central to delivering a great
36:06experience to members. It's impossible
36:08to take the breadth of content that we
36:10have. There's ever more content. That's
36:12one of the challenges we face.
36:14And make discovery easier and easier and
36:17easier, which is one of the challenges
36:19that Netflix has.
36:21And using AI and ML has been a way to do
36:23that. You want to personalize right
36:25title for the right person at the right
36:26moment, that problem gets harder. The
36:29The more exciting our catalog gets, the
36:31greater breadth of content we have, not
36:33just film and TV, but games and live and
36:35podcasts,
36:36personalization becomes even more
36:38important in what that experience is.
36:40So, we can take a lot of that history
36:42and say, "Okay, well, now how do we
36:44solve this problem?" Because the tech is
36:45even more powerful, but it gives us a
36:48running head start in being clear about
36:50the problem to solve, how important it
36:51is that Netflix solve that for our
36:53members. And then the same is true, as I
36:56was mentioning, on the creative side of
36:57the house. AI and ML have been in things
37:00like visual effects or in localizing
37:02language for a long time. Now we say,
37:05"What's the next era of that when the
37:06tech is more powerful?"
37:08And in In both cases, it ends up taking
37:11a strength that Netflix has, which is
37:13marrying entertainment and technology,
37:16and making sure we stay ahead of the
37:17game to deliver things that are even
37:19better. So, I I love that it it's part
37:21of our history. It still continues to be
37:23a strength, and it's going to have to be
37:25a strength, given the size of the
37:26challenges we're facing around the
37:28breadth of entertainment while keeping a
37:30great experience.
37:32>> Yeah. And I I love that back then it was
37:34called machine learning, and AI was
37:36like, "No, no, this It's not AI. AI is
37:38Never never never Never going to happen.
37:40It's just machine learning."
37:41>> Well, then all of a sudden we call
37:43everything AI, and some of it's machine
37:45learning.
37:45>> That's right.
37:45>> So, I
37:46>> [laughter]
37:46>> I I tried to You know, it depends like
37:48the thing that is of the moment to
37:51describe. So, I think we bucket all of
37:53it as AI now.
37:54>> Yeah, AI has become
37:55>> of AI use cases that are not generative
37:58use cases. So, we could go down a deep
38:00dark hole of all the specific things.
38:02But in general, like I don't think it
38:04would surprise anyone that Netflix is
38:05using a broad array.
38:08And it with so much excitement about
38:10what's possible, the fun thing at
38:12Netflix for the people who work here is
38:14that if you're really passionate about
38:16the applications of tech for
38:19creative outlets, for consumer products,
38:21for infrastructure, we have all of those
38:23problems and AI is at the center of them
38:26and it's good not to forget that that
38:28that's true even if Netflix isn't
38:30branded as an AI company. AI is a tool
38:33that we're very comfortable using to get
38:35these great entertainment and technology
38:36outcomes.
38:37>> The other really interesting thing just
38:39to kind of keep complimenting Netflix
38:41here. If you look at the early culture
38:43deck of Netflix and also our
38:46conversation last time,
38:47things that emerge from that are things
38:49like high agency. This is like something
38:51core to Netflix in the beginning. High
38:53agency, autonomy, high talent density,
38:56very bottom-up thinking, super quick
38:59experiments and launching, paying top of
39:02market. Uh
39:04this is all stuff that every AI like
39:06this is what I hear constantly now from
39:07how the top AI labs operate. So we're
39:10all ending here and this is where
39:11Netflix has been forever.
39:13>> Yeah, it's a little prescient in
39:16understanding what makes talent
39:18incredible.
39:20I've thought about all those aspects of
39:21the culture at Netflix as this is going
39:24to sound a little bit nerdy, but
39:25excellence as an operating system.
39:28So the goal of all those cultural
39:30elements wasn't the end goal in
39:32themselves. It wasn't let's just make
39:34sure people have as much responsibility
39:36as possible or let's you know, we don't
39:38like process. So let's make sure that we
39:40don't have any of that.
39:42It was instead a very strongly held
39:45opinion that that you get to excellence
39:48by giving people a lot of agency and
39:50accountability. By pushing decisions as
39:52deep in the organization as possible,
39:55hiring great people who can be trusted
39:57to have good judgment and make good
39:59decisions.
40:00And that ends up driving incredible
40:03outcomes plus a lot more motivation and
40:06sense of responsibility. It means every
40:08person on the team can feel like I'm
40:10being given
40:12a lot of keys and a lot of
40:14accountability for what happens here and
40:16I myself feel like when you know you're
40:18carrying that level of trust and
40:20accountability, you want to do your best
40:22work.
40:24And so there's something that
40:25feels very intuitive about Netflix's
40:28culture has always been aiming at
40:30excellence.
40:31And when you have great talent and you
40:33give them the ability to do their best
40:35work without micromanaging it or
40:37drowning it in process, you actually get
40:39much better outcomes.
40:41And so I do think that the newer era
40:43companies are picking up on something
40:45that is feeling very familiar to us. And
40:48it it's not something that comes easily.
40:49So having culture is not a static thing.
40:52Culture needs to
40:54grow and evolve as a company gets
40:56bigger, the types of problems you're
40:57solving change. But the notion that like
41:00we're going for excellence and trusting
41:02that exceptional talent needs to be able
41:04to do their best work. That's unchanged
41:07and something that I think continues to
41:08be a special sauce for us.
41:10>> I love this concept, excellence as an
41:12operating system.
41:14It's very uh systems thinking, he he
41:16might say,
41:17for how to set up a company.
41:18>> Exactly, Lenny.
41:19>> [laughter]
41:20>> So for people that like everyone
41:22listening to this will want excellence
41:23as an operating system. Like who would
41:25not want this?
41:26Uh it'd be helpful for people to hear
41:27what are kind of the ingredients to make
41:29this happen. One is obviously high
41:31talent density, just hiring only the
41:33best. Two is accountability. Kind of
41:36there's like the input and the output
41:37essentially. Uh input amazing people,
41:39top the top people, give them make them
41:42accountable, give them autonomy. What
41:43would you say kind of like the pillars
41:45of creating this
41:46uh excellence as an operating system if
41:48people if founders are listening to this
41:49like I want them to do that.
41:51>> Well, the talent density is the
41:52non-negotiable. Like you have to start
41:54with that. If you don't have that, you
41:56can't get to a place where you have
41:58confidence in decision-making at all
42:00levels of the organization,
42:03allowing people to take risks and
42:05innovate quickly. That's a big part of
42:07excellence in the Netflix culture, which
42:09is being very comfortable with
42:11risk-taking.
42:12We don't try to avoid failures, we try
42:14to recover quickly when we have them.
42:17I think there's been great examples of
42:19that. Our foray into live was a
42:21wonderful example of being comfortable
42:23taking a ton of risk, knowing it would
42:25be imperfect, knowing we would learn
42:27fast, and we would be better for it.
42:29I've never been prouder of the team
42:30seeing how we worked through that. So,
42:33you have to be talent density,
42:36comfortable that people are going to
42:38take the context that you give them,
42:41strong judgment and risk taking,
42:44and fight for the things that are the
42:46best outcomes for the business.
42:48You have to be very clear that what
42:51you're doing is driving outcomes for
42:53consumers and Netflix.
42:55So, it's Netflix matters, Netflix
42:57members matter. It's not about my own
43:00personal success or what I prefer. So,
43:02there's a selflessness
43:03that is part of this excellence
43:05operating system.
43:08And then the other thing I would say is
43:09some of the things that are they're
43:10really unnatural for humans to do. So, I
43:13could give a couple examples of things
43:15to get comfortable with,
43:17which is
43:18there are certainly days where I see
43:20decisions happening, and I think, "Hmm,
43:24I would make a different decision."
43:26Like, is that really going to be the
43:28best thing?
43:30But, my job, especially in the Netflix
43:32culture, is not to step in in every one
43:35of those cases and overrule or veto or
43:38question someone,
43:40especially if it's
43:42it's not material, it's not going to
43:44burn the place down. Let people make
43:46that decision and learn from it. And ask
43:49for those reflections afterwards of
43:51like, "How did it go? Maybe I was wrong.
43:53Maybe the decision was a great one."
43:55But, that it's related to the risk
43:57taking and the like help people learn
43:59how to feel comfortable making their own
44:01decisions, especially when they're not
44:03all going to be the right decisions, and
44:05they're going to learn something tough
44:07from it. I felt that myself from my boss
44:09and my peers saying, "This is your
44:11decision. You know, I can provide input.
44:13I can help you brainstorm. It's yours in
44:15the end."
44:16And I that it it just doesn't come
44:19naturally when the stakes are high, when
44:20I feel responsible for what the org's
44:22doing to let people lean into risk can
44:24be uncomfortable.
44:26And I think that also means in cases
44:28where things are not going well as
44:30another example to not assume that
44:32process is going to fix it.
44:35So, if
44:36or something I've learned over the past
44:37few years, that
44:39when planning is difficult, I've never
44:41heard someone say like, "Oh, we figured
44:43out the perfect way to plan."
44:45Or the perfect way to go through
44:47feedback and leveling and compensation.
44:51But every time we saw that and we added
44:53more process, we spent more time without
44:56getting better outcomes.
44:58And so, it's another unnatural thing
45:00that I think everyone's inclination when
45:02things are hard and complicated
45:05is
45:06you think you're simplifying the problem
45:08by putting a lot of constraints around
45:10it,
45:11but it actually goes against the like,
45:13is there a more creative way
45:15to plan or to make people decisions or
45:17to make prioritization decisions that
45:20actually get us to better outcomes. And
45:22so, it's a resistance to do the thing
45:24that a lot of bigger companies would do
45:26and to feel comfortable in that
45:28discomfort very often. So, that's
45:31something I feel in my role and I I
45:32would believe a lot of people at Netflix
45:34feel it because you try not to do the
45:36thing that is
45:38standard.
45:39>> It's easy to say that and hear that, but
45:41I so know what you mean, where somebody
45:43screws up and you're like, "Okay, what
45:45was the thing that went wrong? Let's put
45:46a process in place to avoid this from
45:48happening." And what you're saying is
45:49like, you need to resist that.
45:51Uh because that slows things down and
45:54the best people don't want to be working
45:55in a place with all these checklist and
45:57process and gates and things like that.
45:58>> No, I think the best people want to know
46:00there's going to be a blameless retro
46:02and they're going to feel so
46:04individually responsible
46:07that they're going to say, "How do I
46:08make sure this doesn't happen again?"
46:10Not with process, but with like how
46:12could I share these learnings? How could
46:14I do work differently to make sure that
46:16I get to a better outcome next time?
46:19When you are trusting people to take
46:21those reflections
46:23and learn and grow
46:25I think you get much better
46:27outcomes over time. You get a much
46:29stronger team, which I think is part of
46:31our role as leaders of like you're
46:33you're trying to grow a team that is
46:35resilient and durable and knows how to
46:37have great impact. You're not trying to
46:39control everything.
46:41>> Which is a key to building a team with
46:43high talent density.
46:45There's two sides to this that I want to
46:47chat about briefly. One is the hiring
46:48and the other is
46:50keeping the people. So you're famous for
46:52the keepers test. We talked about this
46:53last time. Another unnatural thing for
46:55people.
46:56People that want to understand what this
46:58is, they can listen to the first
46:59conversation, but has that How has that
47:00evolved over the last couple years?
47:02That's still core part of the culture,
47:03this idea of the keepers test?
47:05>> It's often cited in a way where you
47:07think of keepers test as
47:10that moment where you decide to let
47:12someone go, that they're not the right
47:14fit for the role and the conversation
47:16about that.
47:17But it's equally commonly used to have a
47:20conversation about how extraordinary
47:22someone is.
47:24How well they're doing in a role.
47:26Because it the entry point is for me to
47:28say to one of my direct reports or for
47:30them to say to me
47:32"How am I doing on your keeper test?"
47:34And
47:36the lion's share of the time my response
47:38is, "I would fight so hard to keep you."
47:41Let me go through a set of things that I
47:43think you're doing such a great job at,
47:44what your strengths are, where you're
47:46having a lot of impact. Here's how you
47:48could be even better. So it's an entry
47:50into a conversation that is very
47:51positive and uplifting for people, but
47:54the framing is, "Do I pass the keeper
47:55test?" And then, of course, there's the
47:57harder situations where
48:00I'm evaluating does someone pass the
48:01keeper test or they're asking me, and
48:04it's This is the toughest thing to say,
48:06to be honest, you're not passing that
48:08right now.
48:09I think you could get there in some
48:11cases, and that comes with feedback and
48:12what are those milestones? Or in some
48:14cases you're saying, we've really tried
48:16and I don't see the path to success. So,
48:19it it's just it's an anchor and an entry
48:22point for a conversation that can go
48:23lots of different directions. And the
48:26thing I like about it is it's good
48:28hygiene on feedback and checking in on
48:30how things are going
48:32and forcing a tough conversation
48:34sometimes instead of shying away from
48:36it. Or to keep great talent, you do need
48:39to say you're doing great. Like that
48:41that's an important part of making peo-
48:43people feel recognized and valued. So, I
48:45don't want it to come across that we
48:46just have this
48:48very negative view of it. I think
48:50there's this positive side of the coin
48:52as well.
48:53>> Awesome. I guess just to explain to
48:55people what this is so they don't have
48:56to go listen to other podcasts, I'll try
48:58to briefly explain it. The idea here a
49:00part of the Netflix culture is that
49:02when you have people reporting to you,
49:05you should always be thinking, if I were
49:07to would I hire this person today?
49:10Knowing what I know about them, and if
49:11not, then I should probably let them go.
49:13And the idea there is to keep the high
49:15bar, to not ever just like settle, okay,
49:17this person they're here, I guess we'll
49:18keep them around. Is that is that
49:19roughly the way to understand it?
49:20>> Yeah, and the way it it can it's sort of
49:23a corollary to that if that person came
49:26to me today to say they were leaving,
49:27would I fight to keep them or not? Or
49:29would I say, if if my sense is relief
49:32of, oh yeah, it probably would be better
49:34to have someone else in this role, I
49:36should have taken action in having that
49:38conversation sooner.
49:39>> I love
49:40as you said, it's such an so many
49:42uncomfortable things you have to do to
49:43maintain
49:45>> Yeah, it's the Well, the keeper test is
49:48one, maintaining talent entity, context
49:50not control among leaders. We talk about
49:54being highly aligned but loosely
49:55coupled, which is where light process,
49:58you know, the minimum to make sure we're
49:59clear on the priorities and we can
50:01execute them as what we're solving for.
50:03All of these things are not things that
50:05human beings or organizations at scale
50:09tend to do. So, it's constant diligence
50:11to try to maintain the thing that's made
50:13Netflix a special place. Cuz in the end,
50:15it's the work and the culture that
50:17attracts people and retains people, and
50:19we need that to be a successful
50:21business.
50:22>> So, that's exactly where I was going to
50:23go. Uh so, to make this work, you need
50:26to attract the best people. It's always
50:28been very hard to attract the best
50:30people. Feels insanely hard these days
50:32with the amount of dollars flying
50:34around, the fancy AI labs, so much
50:36competition. There's like everyone's
50:38just, you know, it's it's crazy. What
50:40have you found to be uh effective in
50:42convincing the top people to still come
50:44to Netflix and and join versus all the
50:47other fancy places they can go?
50:49>> Yeah, we've always had a lot of
50:50competition for talent. It might feel
50:53more pronounced right now, but we we
50:55have great talent on the team. Maybe
50:57that goes without saying, but I feel
50:58like I should say it out loud cuz I
50:59believe it. We have incredible talent at
51:02Netflix, recent hires, long-tenured
51:05people.
51:06I'm always impressed by the work that
51:08the team is doing. So, I don't feel like
51:12we've suffered or like other companies
51:14are vacuuming up all the good people
51:16because so many of them I do think sit
51:18at Netflix.
51:20It does feel like we have to be more
51:24more explicit about the types of people
51:28and talent that tend to thrive at
51:30Netflix versus other companies like some
51:33of the frontier labs.
51:35So, people at Netflix have to be
51:37passionate about the application of
51:40technology. And the application or
51:42building products to solve a certain set
51:44of problems. You have to love
51:45entertainment. You have to love consumer
51:47products at scale. You have to love the
51:49global nature of that.
51:51There are a lot of incredibly talented
51:53people
51:54who love that sweet spot. I am one of
51:56them between
51:58tech and product and entertainment and
52:00how do you make those things come
52:02together in a way that's remarkable?
52:04And you use AI to do it. You use other
52:06technologies and products to do it. But
52:09that has to be something that drives you
52:11to be really excited about a lot of the
52:13roles at Netflix.
52:15If instead you're by some of the
52:17foundational work that the frontier
52:18model companies are doing, which is
52:20exciting in its own way, it's a
52:22different persona. It's a different like
52:24here's the problem space that I want to
52:25work in.
52:27But I don't think there's a shortage of
52:28people who get really excited about the
52:31applications of the technology and see
52:34the connection to that to things that
52:37they love and use every day like
52:38Netflix. And so that, you know, that
52:40gets me up in the morning and I think it
52:42gets a lot of the team members up and we
52:44have this conversation about like that's
52:46something special that only talent at
52:48Netflix can do or fill in the blank for
52:50another industry that's deep in the
52:51application of it. I think that's
52:53inspiring.
52:54>> I want to kind of touch on a couple
52:56things that I've been thinking about in
52:57this world of AI that we're uh
53:00approaching. One is uh junior people.
53:03It feels like everyone's like there's a
53:05good example. You're hiring a lot of
53:07awesome senior people that have proven
53:08they're awesome and you know, high
53:10talent density, high bars. Uh
53:13also just AI makes it so easy to do
53:15stuff that people may not be learning
53:17how to do anything. They're like junior
53:19engineers I'm thinking or junior PMs,
53:21junior designers.
53:22Like there's just like how do new people
53:25become these awesome senior people? Is
53:28there anything you've
53:30you think about? Are you hiring junior
53:31people? How do you think about this if
53:33this what happens with junior people not
53:35necessarily learning or having a path to
53:37learn to become the senior person?
53:39>> We are still hiring junior people and
53:41they're really important to our talent
53:42strategy.
53:44So, we still have an intern program, we
53:45still have a new grad program, which is
53:47a was new for us as of a few years ago.
53:50So, prior to a few years ago, we were
53:52only hiring more experienced talent
53:54across all the functions. Now, we do
53:56hire people straight from undergrad and
53:59graduate programs and we'll continue to
54:01do that. So, even in a world of AI where
54:04some things are easier, we were talking
54:06earlier about
54:08mindset, AI fluency.
54:12From my experience,
54:14younger folks are more open-minded.
54:17They tend to be more native in some of
54:21these new ways of working. For a company
54:23that like Netflix, they're also very
54:25fluent in how entertainment is changing,
54:27how consumer behaviors are changing, how
54:30product and tech is influencing that in
54:33the products that they're using. That's
54:35really important to have on our team.
54:38So, there there's the part of the
54:39persona, which is who are you as a new
54:41grad who's an engineer, but there's also
54:43who are you as someone who's in their
54:45early 20s and has a perspective on the
54:47world that is highly valuable and a
54:49comfort with the way the world is
54:51changing. So, that's why I say it's a
54:53critical part of our talent strategy.
54:55To the Okay, so you step into the role
54:57and you have AI tools that didn't exist
54:595 or 10 years ago, I would say mastery
55:03of the craft is still very important.
55:05So, going back to as the team member,
55:08I'm responsible for the quality of code
55:10that I am submitting for production, I'm
55:12responsible for the quality of products
55:14that I'm building, how they are
55:16designed, what that user consumer
55:18experience is. None of that is going
55:20away. So, if I think about more junior
55:22or earlier career talent on the teams,
55:25we need to be investing just as much in
55:27the mentorship of this is what good
55:28looks like, this is how you use these
55:30tools, but you still take accountability
55:32for what the outcomes are, what the
55:34quality of the output it
55:36And I think I mentioned this earlier, I
55:37find that mastery and that craft
55:39excellence scarce still. So, we want to
55:41make sure we're teaching that. I I think
55:43it's a valid concern of like, how do I
55:45get that if I'm not as hands-on as I
55:47would have had to be, but you still
55:49carry responsibility for reviewing code,
55:52testing code, being able to diagnose
55:54problems, knowing what a good product
55:57looks like. Like, I think that's a very
55:58scarce skill to say, "This is excellence
56:01in in a product that solves a problem
56:03that matters and in how it's designed."
56:06So, I don't think that craft mastery,
56:08the importance of it, is going away.
56:10Probably the way we train and grow
56:13talent has to change cuz they're going
56:14to use different tools, and I can
56:17guarantee you that earlier career talent
56:20is going to be teaching older folks like
56:22me many new things, too. So, I think it
56:24goes in both directions.
56:26>> Where do you think engineering goes in
56:28the I don't know, 5, 10 years? Do you
56:30think
56:31people need to still understand code? Or
56:35do you think there's this abstraction
56:36layer that sits on top where you don't
56:38even have to learn C++, Java, Python,
56:41whatever?
56:42>> I think there's a difference between
56:44being able to write lines of code in a
56:46particular language like Python or C++
56:50and understanding how code, computer
56:54systems, products work.
56:57And I don't think the latter is going
56:59away.
57:00Because if we trusted agents to know all
57:04the languages and write all the code,
57:05we're not going to know
57:07why is something Is it a good product?
57:10Is it a bad product? Is it working as we
57:11expected when it doesn't? Like I
57:13mentioned earlier, we take a lot of
57:14risk. We fail fast, we recover fast.
57:17That requires an understanding of how
57:20are these systems working. I might use
57:21an agent to help me understand those
57:23things, help me detect an anomaly or
57:26something that's broken faster and
57:27triage it,
57:29but I still need to have a fluency of
57:31like, what is this thing that we're
57:32building and how does it work? So I know
57:34if it's good and I know how to fix it.
57:37I don't know I I hope that doesn't go
57:39away cuz it you know, that that's like a
57:41how do we make the world a better place
57:42through the stuff that we're building? I
57:44think requires some understanding of
57:45what we've built.
57:46>> What I'm hearing which it makes sense is
57:48you may not have to write the code but
57:49you have to understand it and what's
57:50happening. But it's so much harder to
57:53just as a person not writing it to
57:55actually you know, have that instilled
57:58in you.
57:59>> I think that's one of the the things
58:00that the learning curve is very steep on
58:02right now. So looking at some of the
58:04code that some of these models or agents
58:07are writing
58:08they're very hard to follow.
58:10It's like I know I'm getting better
58:11performance from this but I have no idea
58:13why and if this thing breaks I'm going
58:15to have no idea how to fix it.
58:17That that's makes me uncomfortable. You
58:19know, maybe that's because I'm still on
58:21that learning curve of like how do we
58:22operate in that world? Like what's the
58:24set of tests or rationalization and
58:27understanding that we need to have to
58:28get comfortable with it? But at first
58:30glance it looks very unfamiliar and very
58:33unsettling. So I think engineering over
58:36time will evolve to be comfortable with
58:38that and have fluency in it and know how
58:40to guide new tech and agents and new
58:43capabilities to make sure that we feel
58:46really good about what the output is.
58:48>> I wonder what the metaphor is for this
58:49where this like it's I continue to be
58:51astounded by how much engineering has
58:53transformed in like 2 years. It's like a
58:56completely different drop down. You're
58:58just used to sit there and then I would
59:00write code and now you're just
59:02talking to agents and reviewing code and
59:03shipping a bunch PRs a day.
59:05>> It feels like it's a it's an
59:07acceleration of how much engineering has
59:10changed. But if you looked over the last
59:1310 years or 20 years you would say the
59:15same thing.
59:16>> Mhm.
59:17>> So it there's just something that's
59:19moving faster and it's hard to wrap our
59:22heads around how quickly it's moved in
59:24the past couple of years but it's not
59:27it's not totally unfamiliar that
59:29engineering or data science or product
59:32would have these big shifts, just like
59:35how filmmaking works. If you will go
59:37over the last 100 years, it's
59:39unbelievably different because of
59:41technology and new tools that we brought
59:43to it.
59:44Just feels like the cycle is speeding
59:45up.
59:46>> Okay, I want to talk about entertainment
59:48for for a brief moment. Just
59:50I'm curious just like how entertainment
59:52will change over time and say like 5, I
59:55don't know, 5, 10 just you know, today
59:57we open up Netflix, check out some
59:59shows, watch some videos. It hasn't
1:00:00changed in a while, just that idea of
1:00:02like cool, I'm going to watch the pit
1:00:04and watch it all. I'm going to watch a
1:00:05movie. Uh I got TikTok, I got Instagram
1:00:07feeds of stuff. Like how much different
1:00:10do you think this will be in I don't
1:00:11know, 5 years?
1:00:12The way we entertain ourselves.
1:00:13>> it's already changing at Netflix
1:00:16because entertainment is not going to be
1:00:18one thing in the future and it's already
1:00:20not one thing now. So, part of the
1:00:22reason that we are going beyond film and
1:00:25TV in our offering
1:00:27is because there's there's an
1:00:28expectation that consumers have of much
1:00:31greater variety across formats, devices,
1:00:34moments of the day that Netflix needs to
1:00:37be able to serve well in order to meet
1:00:40consumer expectations and hopefully
1:00:42exceed them over time. So, when we think
1:00:44about the addition of mobile and TV or
1:00:47cloud games
1:00:49live content podcasts, working with a
1:00:52broader set of creators who are now on
1:00:54the Netflix service
1:00:56all of those things create a greater
1:00:59breadth of what entertainment is and
1:01:01Netflix is able to define and expand
1:01:03that.
1:01:04And it puts a higher bar expectation on
1:01:07how do we make sense of that for a
1:01:09Netflix member?
1:01:10So, how do we show you this very
1:01:13seamless journey from I listen to the
1:01:15Bill Simmons podcast to I watch
1:01:18Quarterback because I love that as one
1:01:20of the Netflix offerings in the more,
1:01:23you could say, traditional film or TV
1:01:24space
1:01:25to I play the most recent FIFA cloud
1:01:29game.
1:01:30And I want to be able to do that in both
1:01:32TV and on my mobile phone because now
1:01:34I'm on the move and I want to be able to
1:01:36discover and engage with the content at
1:01:38different moments of the day.
1:01:40That's already a journey that we're
1:01:41building into Netflix, which I think
1:01:43will become stronger and stronger over
1:01:45time.
1:01:45So, the future of entertainment isn't
1:01:47going to be one thing and it's going to
1:01:48have to be more personalized, more
1:01:50immersive, more interactive with this
1:01:53sense of this is a world that I can
1:01:55explore in lots of different directions
1:01:57depending on what I'm looking for in the
1:01:59moment. And that the challenge Netflix
1:02:01has is we've got to make discovery and
1:02:04engagement much easier than it feels
1:02:06today. We have tons of content and it
1:02:07can feel very fragmented, especially
1:02:09when you consider all the services or
1:02:11offerings out there.
1:02:13And I I think Netflix is very well
1:02:14positioned to understand how to solve
1:02:16that problem across entertainment,
1:02:18product, and tech.
1:02:19>> The other element of this is AI,
1:02:21obviously. As an outside observer, it's
1:02:24like so interesting to see how in tech,
1:02:26it's like AI, I love it. It's the
1:02:27future. It's the best. In Hollywood,
1:02:29it's like, "No. Shut it down." There's
1:02:33>> There's a mix. There's a very wide
1:02:35array. So, we Netflix's role in this is
1:02:38to enable creators with whatever tools
1:02:41they want to use to bring their vision
1:02:43to life.
1:02:44There are going to be some creators or
1:02:46filmmakers who are on the end of the
1:02:47spectrum that says, "Absolutely not. No
1:02:50AI. That is not how I do production.
1:02:52It's not It's not consistent with my
1:02:54vision."
1:02:55That's fine. We work with those
1:02:57creators.
1:02:58There's other creators
1:03:00a growing number of them, I would say,
1:03:01who are very interested in exploring,
1:03:04"Wait, can these gen AI tools make
1:03:05something possible that wasn't possible
1:03:07before?
1:03:08Can I tell a story in a new way? Can I
1:03:11make that story higher quality and more
1:03:13resonant for audiences? Can I do things
1:03:16that are extra creative and how I think
1:03:18about bringing a story to life?
1:03:20And we support them as well, and we
1:03:21support all the folks who are in the
1:03:23in-between. And then that's a really
1:03:25important position for us to be in
1:03:27again, because entertainment is not
1:03:28going to be one thing. There's not going
1:03:30to be one format. I think there's going
1:03:32to be types of film and TV that feel
1:03:34traditional, and then there's going to
1:03:35be entirely new formats that
1:03:38unbelievable creators help to bring to
1:03:39life, and Netflix wants to participate
1:03:41in that. Which means we need to have a
1:03:43flexibility in the tools that we provide
1:03:46and the types of partnerships we have,
1:03:48and to really have a creator enablement
1:03:50view rather than a prescriptive that we
1:03:53only do this one way.
1:03:54>> I think people are going to be surprised
1:03:55by just how good AI content is. Like
1:03:58Spencer Pratt's videos are just like
1:04:00everyone's like, "Wow, this is
1:04:01entertaining." Obviously AI, but it's so
1:04:03interesting. Do you think Do you think
1:04:05we'll get to a place where it's just
1:04:06like whole TV shows are AI and people
1:04:08love it?
1:04:09>> I have a hard time picturing
1:04:10entertainment that doesn't have humans
1:04:12at the heart of it. So that that's
1:04:15humans in the creation of the
1:04:16storytelling, which I think is
1:04:19a scarce and valuable skill. Yeah,
1:04:22storytelling is
1:04:24one and the same with humanity.
1:04:26And like knowing what connects with
1:04:28people.
1:04:29So I think humans will be part of the
1:04:32always be a core part or a critical part
1:04:34of the story.
1:04:36And I think
1:04:37watching
1:04:38watching characters on screen who don't
1:04:41have that humanity
1:04:44feels less compelling to me.
1:04:46And what the power of storytelling
1:04:48really is, to like see another human and
1:04:51to watch how they perform a role or like
1:04:54bring an emotion to life. That's such a
1:04:56human element. Will AI help to bring
1:04:58that to life?
1:05:00Will play a material part in some of
1:05:02those productions or how we get them to
1:05:04look and feel a certain way?
1:05:06Yeah, definitely.
1:05:08But I don't see the version of it that
1:05:10doesn't have the human as the backbone.
1:05:12>> There's a quote that I think is
1:05:14misattributed to Salman Rushdie, which
1:05:17is when a child is born, they first ask
1:05:20for food and water and projection, and
1:05:24then they ask for tell me a story.
1:05:27>> It's a thing going back
1:05:30since the beginning of time
1:05:32that storytelling has been a key part of
1:05:35community and social networks and human
1:05:38feeling and connection.
1:05:40So, I love the idea that technology can
1:05:43amplify that and can bring that to life
1:05:45in very new, novel, exciting ways.
1:05:49But, if
1:05:50to say storytelling wouldn't have that
1:05:52humanity at the center feels
1:05:55like something would be missing.
1:05:56>> Mhm. We're going to see some wild
1:05:58over the years coming out of this.
1:05:59>> Oh, I'm sure. There's no question about
1:06:00that. And a lot of it could be very
1:06:02entertaining.
1:06:04You know,
1:06:05I I don't debate that, either. But, I
1:06:07think there's going to be a broad range,
1:06:09and I think Netflix needs to be at the
1:06:11center of shaping that and bringing that
1:06:13to life, which is our plan.
1:06:15>> Amazing.
1:06:16Well, we covered a lot of ground,
1:06:18Elizabeth. Uh before we get to our very
1:06:20exciting lightning round, is there
1:06:21anything else that you wanted to share,
1:06:23leave listeners with, maybe double down
1:06:25on from things we've talked about?
1:06:27>> It probably came across throughout, but
1:06:28I I would underscore that this is a
1:06:30really exciting time to be building
1:06:32products in entertainment.
1:06:34Everything we talked about of like
1:06:35what's changing in the tech and
1:06:37consumers and like what is entertainment
1:06:40we're at this unbelievable high-velocity
1:06:44innovation period. So, it's what keeps
1:06:45me at Netflix. I think it's a fun place
1:06:47to be. I would be missing something if I
1:06:50didn't reinforce that I think that's
1:06:52true.
1:06:53Um I also think that as an industry we
1:06:55spend a lot of time sometimes talking
1:06:57about the the pure tech or the the
1:07:00capability
1:07:02and we sort of lose the forest for the
1:07:03trees. We're trying to build great
1:07:04consumer products that people love.
1:07:06We're trying to make great entertainment
1:07:08that people love. And it's their
1:07:10favorite thing that I don't want that to
1:07:13be lost in Of course, there's amazing
1:07:15tech and product stuff that sits
1:07:17underneath, but in the end, the thing
1:07:18that's most inspirational is what do we
1:07:21bring to people around the world?
1:07:23>> And on those lines, there's been such a
1:07:25uh
1:07:26the opposite of glut, a drought of
1:07:28consumer new consumer products, consumer
1:07:30experiences. Like there's very few
1:07:32success, like almost no consumer startup
1:07:34works. Uh and AI feels like an
1:07:37opportunity for something else to work
1:07:39and I feel like Netflix is one of the
1:07:40rare companies and brands that continues
1:07:42to deliver an awesome consumer product
1:07:44and business. There's just not that many
1:07:45of them.
1:07:46>> Yeah. We're going to keep that up.
1:07:49>> Well, with that, we reached our very
1:07:51exciting lightning round. We've got five
1:07:52questions for you. Are you ready?
1:07:54>> Okay, I'm ready.
1:07:55>> All right. What are two or three books
1:07:57that you find yourself recommending most
1:07:59to other people?
1:08:01>> I mean, I have to come up with different
1:08:03books than I said last time.
1:08:04>> I don't know. But I I think that sounds
1:08:06great.
1:08:06>> I still like a good throwback. So, two
1:08:09that are coming to my mind
1:08:12Into Thin Air,
1:08:13Jon Krakauer,
1:08:15and Liar's Poker, Michael Lewis. So, I I
1:08:18worked on Wall Street and I like
1:08:20reminding people what it was like in the
1:08:22way back time.
1:08:23>> Favorite recent movie or TV show you
1:08:25really enjoyed, which is maybe too hard
1:08:27for someone working at Netflix, but I'm
1:08:28going to see what comes out.
1:08:29>> The The list is very long. Um the most
1:08:31recent I watched, Remarkably Bright
1:08:34Creatures, after a recommendation from
1:08:36my mom. It's a tearjerker. Talk about
1:08:38the human part of storytelling.
1:08:41>> Favorite product you've recently
1:08:43discovered that you really love?
1:08:44>> Critical for my health and well-being,
1:08:46Eight Sleep.
1:08:48>> Do you have a favorite life motto that
1:08:51you often come back to in work or in
1:08:52life?
1:08:53>> I often go back to the things that my
1:08:55parents instilled in me in very early
1:08:58times. So,
1:09:00the the risk of repeating, maybe.
1:09:03First, something good happens every day.
1:09:06Watch for it.
1:09:08Even in the most stressful times.
1:09:10And second, that the last 5% of effort
1:09:14usually makes all the difference.
1:09:17>> These are awesome. I They hit They hit
1:09:19me.
1:09:20Final question. I don't know what
1:09:23anything about this, but you mentioned
1:09:24you're doing some kind of cycling event.
1:09:26>> Oh, yeah.
1:09:27>> Tell us Tell us what's going on. What
1:09:28are you doing here?
1:09:30>> So, my husband and I are doing a trip
1:09:32where we ride alongside the Tour de
1:09:35France for the last week of the race.
1:09:39So, the tour is 3 weeks. The last week
1:09:41has a lot of mountain stages. So, we get
1:09:43to ride part of the route each morning
1:09:46and then watch the race in the
1:09:48afternoon.
1:09:49Not for the faint of heart. So, I'm
1:09:51trying to train up so I can enjoy those
1:09:53rides. It's supposed to be vacation
1:09:55after all.
1:09:56>> [laughter]
1:09:57>> My god. I love this vacation. We're just
1:09:59going to a race.
1:10:00>> I love cycling. I love professional
1:10:02sports. It's fun to be able to
1:10:04participate in it.
1:10:06>> So, is this like racing or you just kind
1:10:07of try to go as
1:10:09nonchalantly through the course?
1:10:11>> You go nonchalantly. But, still there
1:10:12are I think I mentioned Yeah, it is It's
1:10:15physically and mentally challenging. And
1:10:18I you know, it's not a race, but I don't
1:10:20want to be at the back of the pack. So,
1:10:22I got to be comfortable enough to hold
1:10:24my own.
1:10:25>> Wow. I love how different this is from
1:10:27your job. And it feels like something
1:10:29else to do.
1:10:29>> It's a good balance and it gets me
1:10:32outdoors and gives me some nice
1:10:34perspective. So, I'm looking forward to
1:10:35it.
1:10:36>> Elizabeth, you are awesome. Two final
1:10:37questions. Where can folks find you
1:10:39online if they want to follow you, reach
1:10:41out for maybe anything that came up? And
1:10:43how can listeners be useful to you?
1:10:45>> The best place to find me and some of
1:10:47the work we're doing or reach out is the
1:10:49Netflix tech blog, actually, where we're
1:10:51putting a lot of things that I've been
1:10:52talking about up there. We're trying to
1:10:54do a better job communicating about the
1:10:55fun stuff we're working on. So, that's a
1:10:58good first stop, usually. Um and then
1:11:02how listeners can be useful,
1:11:04try all the new stuff that we're putting
1:11:06out there.
1:11:07Um watch the live events, play the
1:11:10games, have fun with the new vertical
1:11:12video feed that we have on mobile called
1:11:14Clips, send us feedback. So, we want to
1:11:17make it better, and a lot of these
1:11:19things are new zero-to-one efforts for
1:11:21us. So, we're trying to get to great and
1:11:24excellent as quickly as possible.
1:11:25>> I love that the homework is go watch
1:11:27Netflix and
1:11:29>> You can also watch other things, tell us
1:11:30how we can be better, but I'm I'm
1:11:32definitely interested in how can we be
1:11:34better at Netflix.
1:11:36>> I love it. I'm going to go I'm going to
1:11:37go do that. Elizabeth, thank you so much
1:11:39for being here and being here again.
1:11:41>> Thank you for having me. Always fun.
1:11:44>> Thank you so much for listening. If you
1:11:46found this valuable, you can subscribe
1:11:47to the show on Apple Podcasts, Spotify,
1:11:50or your favorite podcast app.
1:11:52Also, please consider giving us a rating
1:11:54or leaving a review, as that really
1:11:56helps other listeners find the podcast.
1:11:58You can find all past episodes or learn
1:12:00more about the show at
1:12:02lennyspodcast.com.
1:12:04See you in the next episode.

About this transcript

This page contains the full transcript of Why Netflix is betting on systems thinkers—not specialists—in the AI era | Elizabeth Stone (CPTO) by Lenny's Podcast, generated from the public captions YouTube serves with the video. The transcript has 13,484 words across 2,115 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.