Gaurav Misra (CEO and co-founder of Captions)27 March 2025

How to win in the AI era: Ship a feature every week, embrace technical debt, ruthlessly cut scope, and create magic your competitors can't copy

6Frameworks

15Insights

Episode overview

Lenny Rachitsky speaks with Captions co-founder and CEO Gaurav Misra about operating a fast-moving AI startup, including the goal that every engineer ship a marketable feature each week. Misra explains how ruthless scope reduction preserves quality, why startups can deliberately incur technical debt, and how Captions separates customer-requested work from a secret roadmap of behavior-changing bets. Drawing on his experience at Snap, he also discusses design-led product development, rapid prototyping, AI video’s evolution, and the risks of synthetic media becoming indistinguishable from reality.

Key ideas

Require every engineer to ship a marketable product or feature each week, then double down on the releases that attract real use.
Protect quality by cutting scope repeatedly until removing anything further would make the product useless.
Treat technical debt as strategic leverage for a startup, while monitoring the maintenance interest that gradually consumes engineering capacity.
Maintain both a public roadmap of requested, competitive-table-stakes features and a secret roadmap of unrequested, potentially behavior-changing ideas.
Use company-wide brainstorming, social trends, prototypes, and user reactions to identify demand and prioritize new AI capabilities.
Build cross-functional product teams whose designers, engineers, and product managers understand one another’s crafts and connect product decisions to marketing.
Focus generative video on storytelling rather than fabricated documentation, and design products to discourage deceptive representations of reality.

Transcript available · source text is retained privately and is not published

Frameworks in this episode

Productivity3 steps

Cut Scope, Not Quality

When time is short, strip features down to the core instead of shipping the same feature badly

Read →

Innovation3 steps

Documentation vs Storytelling Safety Framework

Sort every AI-video use case into documenting reality (block) or storytelling (enable)

Read →

Leadership4 steps

Internal Virality for Alignment

Get a big org aligned by making a prototype go viral inside the company instead of grinding stakeholder meetings

Read →

Strategy4 steps

Strategic Technical Debt as Leverage

Startups should take on technical debt on purpose — it's how you outrun bigger companies

Read →

Strategy4 steps

The Public and Secret Roadmap

Split your roadmap into what users ask for and what only you can see — the wins come from the second

Read →

Innovation5 steps

The Weekly Marketable Feature

Every engineer ships one feature per week that a user would pay or show up just for

Read →

People & resources mentioned

Attributed to the moment in the episode. Timestamps are approximate.

People · 6

Gaurav MisraMentionsgorov was an early employee at snap where he led the design engineering team
“today my guest is gorov misra”
▶ Watch at ~1:00approx.
Lenny RachitskyMentionstoday we've got another very special compilation episode something I've been pulling on more and more with the podcast and the newsletter
“gorov thank you so much for being here and welcome to the podcast”
▶ Watch at ~4:30approx.
Evan SpiegelMentionshe's also VP of product at snap where he's the right-hand man to Evan Spiegel for 2 years
“the CEO Evan Spiegel was that he had an unmatched understanding of the user”
▶ Watch at ~33:30approx.
Patrick CollisonMentions“there's this tweet by Patrick hollison that I've brought up a couple times on this podcast”
“so Patrick hson once tweeted this tweet that really stuck with me”
▶ Watch at ~55:30approx.
Will SmithMentionsmentioned
“how long has it been since that Will Smith spaghetti video”
▶ Watch at ~1:09:30approx.
Elon MuskMentionsWhether you like him or not, Elon, this is similar to his um if you watch him, he has sets these really ambitious goals
“Elon did it well with that we've reached our very exciting lightning round”
▶ Watch at ~1:20:00approx.

Resources · 33

Snap Inc.Mentionscompany
“gorov was an early employee at snap where he led the design engineering team”
▶ Watch at ~1:00approx.
MicrosoftMentionscompany · Bill Gates and Paul Allen
“he's also an engineer at Microsoft and a couple other companies”
▶ Watch at ~1:00approx.
Lenny's NewsletterCoinednewsletter · Lenny Rachitsky
“I just launched an insane deal for subscribers of my newsletter”
▶ Watch at ~2:00approx.
BrexMentionscompany
“this episode is brought to you by brex the financial stack used by one in every three us Venture back startups”
▶ Watch at ~2:30approx.
ParagonMentionssoftware · Brandon Foo
“this episode has brought to you by Paragon the integration infrastructure for B2B SAS companies”
▶ Watch at ~3:30approx.
ActionKitMentionsproduct · Paragon
“leverage action kit their latest product to instantly give your AI agents access to over 100 Integrations”
▶ Watch at ~4:00approx.
Iron ManMentionsfilm
“you've seen Iron Man and stuff like they have the Jarvis thing”
▶ Watch at ~8:30approx.
InterstellarMentionsfilm · Christopher Nolan
“you've seen like U Interstellar and they have like the tar machine”
▶ Watch at ~8:30approx.
CursorUsessoftware · Anysphere
“we are using cursor yep”
▶ Watch at ~24:30approx.
DevinUsessoftware · Cognition AI
“we're using Devon as well which is another you know that's more advanced”
▶ Watch at ~24:30approx.link ↗
Eye ContactCoinedproduct · Captions
“one of the first sort of AI features he added after sort of the app initially took off was this feature called ey contact”
▶ Watch at ~30:00approx.
NvidiaMentionscompany
“we worked with Nvidia on this”
▶ Watch at ~30:30approx.
InstagramMentionssoftware · Meta
“you're not going to open Instagram or anything else because it doesn't open to the camera”
▶ Watch at ~34:30approx.
SnapchatUsessoftware · Snap Inc.
“you're going to open SnapChat right because you can capture it right away”
▶ Watch at ~34:30approx.
TikTokMentionssoftware · ByteDance
“other than Tik Tok which I don't think is a social network”
▶ Watch at ~35:00approx.
FacebookMentionscompany · Meta
“there's a lot of reasons for Facebook or any other social network to stop the growth of Snapchat”
▶ Watch at ~37:00approx.
AirbnbMentionscompany · Brian Chesky, Joe Gebbia, and Nathan Blecharczyk
“Airbnb is a consumer product it's not a social network”
▶ Watch at ~40:00approx.
CodaUsessoftware · Coda
“I personally use Kota every single day to manage my podcast and also to manage my community”
▶ Watch at ~48:00approx.
Lenny's PodcastCoinedpodcast · Lenny Rachitsky
“I personally use Kota every single day to manage my podcast”
▶ Watch at ~48:00approx.
StripeMentionscompany · Patrick Collison and John Collison
“like stripe I think they had hundreds of Engineers before they hired the first PM”
▶ Watch at ~51:00approx.
SpotlightMentionsproduct · Snap Inc.
“even on Spotlight the Big Challenge was like how do you take something like that and put that inside the snap Mission”
▶ Watch at ~58:00approx.
Our StoriesMentionsproduct · Snap Inc.
“there was this product called our stories”
▶ Watch at ~59:00approx.
ChatGPTUsessoftware · OpenAI
“my parents live in India and they are the only ones in the neighborhood that know about chat GPT”
▶ Watch at ~1:11:00approx.
CaptionsCoinedcompany · Gaurav Misra and his co-founder
“the first thing we did was build the captions app”
▶ Watch at ~1:15:00approx.
XMentionswebsite · X Corp.
“I mean I think X is actually a great example of that too”
▶ Watch at ~1:19:30approx.
Silo seriesMentionsbook · Hugh Howey
“there's a book uh around Silo too I read that I read all of them there's three of them”
▶ Watch at ~1:21:00approx.
SeveranceMentionstv show · Dan Erickson
“I like Silo and sance”
▶ Watch at ~1:21:00approx.
SiloMentionstv show · Hugh Howey
“I like Silo and sance”
▶ Watch at ~1:21:00approx.
LinearRecommendssoftware · Linear
“my favorite product honestly is linear I'm not going to lie just because it's so welld designed and it's so easy to use”
▶ Watch at ~1:21:30approx.
NotionMentionssoftware · Notion Labs
“a year you're free of linear and superhuman and notion and granola”
▶ Watch at ~1:22:00approx.
GranolaMentionssoftware · Granola
“notion and granola which is incredible AI app for not taking and perplexity perplexity Pro”
▶ Watch at ~1:22:00approx.
SuperhumanRecommendssoftware · Superhuman
“I also like superhuman I mean these are obvious answers but I do use these things every day”
▶ Watch at ~1:22:00approx.
Perplexity ProMentionssoftware · Perplexity AI
“perplexity perplexity Pro nice $2,000 in value for the price of my newsletter”
▶ Watch at ~1:22:00approx.

Spot an error or want something removed? Request a correction or removal.

Insights & moments

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

Hot Take· 7

Hot Take05:30

Why Right Now Is a Once-in-a-Career Window to Start a Company

Misra argues this is a uniquely rare moment in tech where ideas that felt done or already-tried suddenly work. There are more solvable problems than there are people to work on them, and he warns the window won't last forever.

Five to seven years ago starting a company felt impossible because everything seemed already done or already tried and failed
Now the technology lets you build things that were never possible before, so 'everything you try just works'
There are more things that can be done than there are people available to do them
This condition is temporary and the industry will eventually catch up to it

“everything you try just works there's so many possibilities there's not enough people in in the world to work on them”
Gaurav Misra · 05:30

#startups#ai#timing#opportunity

Hot Take07:30

A Playground People Leave Isn't a Business

With building now easy, the hard part is getting people to pay attention and stay. You can pull people in by saying you 'rethought this with AI,' but if you're just a cool playground people try and abandon, you don't have a business. You still have to solve a real problem.

The tech to build is now trivial; the scarce thing is attention and retention
Saying 'we rethought this with AI' will get people to try your product, but only delivering keeps them
A product people play with and then leave is not a business
Solving a real problem is still the core requirement

“if you're just building a playground and people play around in the playground and then they leave after playing around it's not a business right”
Gaurav Misra · 08:00

#product#ai#retention#attention

Hot Take20:30

A Startup's Job Is to Take On Technical Debt

Misra reframes technical debt as strategic leverage, like financial debt used to buy a house you couldn't otherwise afford. A startup should push problems to a future 50th, 100th, or 500th engineer, because if the company fails that engineer is never hired anyway. The catch: every debt charges daily interest, and enough of it means you run out of a 'technical debt runway.'

Bigger companies avoid or repay technical debt; startups should deliberately take it on to move faster
Ask whether a problem must be solved today or can be left to a future engineer who doesn't exist yet
Technical debt is like financial leverage: it lets a small team build what it otherwise couldn't
Every piece of debt costs 1-2% of your time daily in maintenance; too much and you spend all your time 'keeping the lights on'

“as a startup your job is to take on technical debt right because that is how you operate faster than a bigger company”
Gaurav Misra · 20:30

“every piece of debt that you take on you have to pay interest on right so if there is Deb that you've taken on there's…”
Gaurav Misra · 22:30

#technical-debt#startups#engineering#leverage

Hot Take52:30

PMs Should Own the Product All the Way to Marketing

Misra argues marketing is just expanding the surface area of the product, placing a button to your app inside Google or Facebook. The user journey and every metric begins the moment someone taps that ad, so understanding users from that point on is the same skill set as product. He thinks PMs should own the funnel all the way from the ad click.

The biggest opportunity for PMs is understanding how to find the users who have the problem
Marketing is expanding the product's surface area: a search ad or Facebook ad is just a button to your app
The funnel and all its metrics begin the moment a user taps the ad, so it's really product work
Every function should understand the others deeply, ideally well enough to operate in them

“I actually think PMS should own all the way to Marketing in a way and the reason is that if you think about marketing it's…”
Gaurav Misra · 53:00

#product-management#marketing#growth#funnel

Hot Take1:02:30

AI Video Could Return Us to the 1800s

Misra points out that a world where images, audio, and video can't be trusted isn't new, it's how things worked before recording technology existed. If everything can be generated and looks perfectly real, we may return to an 1800s-style world where nothing can be proven except by physical evidence or 'he said, she said.'

Before the 1800s there was no video, audio, or images, and disputes were settled by 'he said, she said'
If generated media becomes indistinguishable from real, that old world could return
Proof would again rest on physical evidence or testimony
Fully indistinguishable video is still a couple of years away, but very close

“there was a world in the 1800s where there was no video or audio or or images right and everything was proven by he said…”
Gaurav Misra · 1:02:30

“we might actually return to that world right where there's no way to prove anything besides you know physical evidence or he said she said”
Gaurav Misra · 1:03:00

#ai-video#trust#deepfakes#society

Hot Take1:20:30

Why He Deliberately Decided Not to Read Books

Asked for book recommendations, Misra reveals he doesn't read books, and it's a deliberate choice. He decided not to build his skill in reading and instead to build it in listening and watching, because he thinks that's the future.

Misra doesn't read books and made that a conscious decision
He deliberately chose not to develop his reading skill
He is instead building his skill in listening and watching
His reasoning is that listening and watching are the future

“I actually don't read books it's actually something that I decided on purpose where I decided I don't want to build my skill in reading…”
Gaurav Misra · 1:20:30

#learning#media#contrarian#habits

Hot Take1:22:30

The Easiest Way to Be the Best Is to Be First

Misra shares the motto that guides how he operates at work. He loves to compete and win, and believes winning means being the best, but the easiest path to being the best is simply to be first.

Misra frames work around competing and winning
To win, you have to be the best
The easiest way to be the best is to be first
It's a personal operating framework someone else pointed out he keeps repeating

“to win you have to be the best but I also think the easiest way to be the best is to be the first”
Gaurav Misra · 1:23:00

#competition#motto#first-mover#mindset

Explainer· 3

Explainer26:30

The Secret Roadmap Competitors Can't Copy

Captions splits its roadmap in two. The public roadmap is everything users have requested, which every competitor already knows and is building. The secret roadmap is things nobody asked for, born from a unique understanding of the problem, and it's where the game-changing wins come from.

The public roadmap is user-requested features every competitor sees and prioritizes similarly
Winning on the public roadmap only gets you marginal gains in prioritization or execution
The secret roadmap holds ideas nobody asked for, kept private until launch
The biggest, game-changing wins come from the secret roadmap, not user requests

“if a user is asking us for it they're asking everybody for it and every team has essentially more or less the same list”
Gaurav Misra · 27:30

“we have a second road map which we think of as a secret road map so this is a road map that nobody asked for…”
Gaurav Misra · 28:00

#roadmap#strategy#competition#product

Explainer38:00

How Snap Ran With Designer-PMs and Almost No PMs

For years Snap had essentially no PMs; a tiny design team of 10-12 people ran product even at 5,000-6,000 employees. These weren't average designers, they were designers who also did the PM job, and they were IC designers with no reports. Because every UI change had to go through this small group, the CEO kept granular control over everything user-facing.

Snap had no PMs for a long time; they came in only with monetization
A design team of just 10-12 people ran product at a company of 5,000-6,000 employees
The designers were IC contributors with no reports who also owned the PM work, roadmap, and shipping
Funneling all UI through a handful of designers gave the CEO granular control over every user-facing change

“there's no PMS really like for a long time for a long time initially there were no PMS at all PMS were introduced with monetization”
Gaurav Misra · 38:00

“these weren't your sort of average designers right these were designers who were actually PMS as well that's kind of what the secret sauce was”
Gaurav Misra · 38:30

#snap#design#product-management#org-design

Explainer1:06:30

The Documentation vs Storytelling Line for Safe AI Video

Captions divides all video into two categories to guide its safety design. 'Documentation' captures what actually happened, personal memories or a reporter recording a real event, and generating fake versions of it is purely negative. 'Storytelling', like ads, movies, and social posts, is understood to be fabricated entertainment, so enabling it is pure positive. The product design deliberately makes one use easy and the other hard.

Documentation video records reality: personal memories, news, history, where fakes add nothing and only harm
Storytelling video is ads, TV, movies, and social posts that everyone knows are fabricated for entertainment
Nobody thinks the Geico gecko is real, so storytelling generation is a pure positive
Captions designs its products to make the storytelling use easy and the documentation-faking use hard

“if we are generating fake versions of reality to full people like there's just nothing good about that right and we want to stay away…”
Gaurav Misra · 1:07:30

“nobody believes like if you watch a guo commercial right like you're not thinking that the gecko is real selling Insurance somewhere out there”
Gaurav Misra · 1:08:00

#ai-video#safety#product-design#ethics

Story· 4

Story30:00

The Eye Contact Feature That Nvidia Didn't Get

One of Captions' first AI features shifts a speaker's eyes to look at the camera so it doesn't look like they're reading a script. Captions built it first, working with Nvidia, who were initially unsure why anyone would want it. The demo went so viral it was recreated in nearly every language and is still copied across apps today.

The eye contact feature makes someone reading off-screen look like they're looking at the camera
Captions was the first company to build it, partnering with Nvidia
Nvidia didn't understand the use case and shared the technology openly
The launch social post was so viral it spread worldwide and has been copied across essentially every app

“we had this feature where it basically shifts her eyes to look at the camera and we were actually the first company to build this…”
Gaurav Misra · 30:30

#ai-video#captions#virality#nvidia

Story34:00

Why Snapchat Declared Itself a Camera Company

Misra explains the mocked decision to call Snap a camera company. The real insight: Snapchat opens directly to the camera, so when a funny moment happens you reach for Snap, not Instagram. Instagram can never copy this because opening to the camera would tank all their metrics.

People laughed when Snap declared itself a camera company
The differentiator was that Snapchat opens straight to the camera to capture fleeting moments
That single small decision held the entire company against all competition
Instagram can't copy it because opening to the camera would collapse their metrics

“Snapchat opens to the camera and that was actually the differentiator right that was actually that small decision was holding the entire company against all…”
Gaurav Misra · 34:00

“Instagram can never copy that because all their metrics are going to go down as soon as they do that”
Gaurav Misra · 34:30

#snap#product-strategy#differentiation#moats

Story46:30

Creating Alignment by Making Ideas Go Viral Inside the Company

Misra's design-engineering team at Snap discovered they could skip the usual stakeholder alignment grind by building prototype products and sharing the builds internally. Interesting builds spread person to person like consumer virality, until engineers, managers, VPs, and eventually Evan were all talking about them, producing instant company-wide alignment.

Big-company alignment is normally slow, done by PMs talking to every stakeholder
A large enough company behaves like a consumer base: share something interesting and it spreads
The team built prototype products and shared the builds to trigger internal virality
Excitement climbed from engineers to VPs to the CEO, creating instant alignment and pull to ship

“you can actually create alignment by causing internal virality if there's enough people in the company it actually starts acting like a like a consumer…”
Gaurav Misra · 46:30

“it would explode like it would just go viral inside the company like day after day we would hear from you know engineers then managers…”
Gaurav Misra · 47:30

#alignment#prototyping#snap#org-design

Story1:14:30

Captions Made $500K On Its Own While They Ignored It

The Captions app took two days to build and immediately hit the top of the App Store, but the founders assumed success couldn't be that fast and spent a year and a half building social networks instead. Sitting on Misra's personal account, the app quietly grew to $500K in revenue with no employees, releases, or support, and 2,000 unanswered tickets. When they finally focused on it, growth went so vertical the old curve looked flat.

The Captions app took two days to build and shot to the top of the App Store on its own
The founders got distracted and spent ~18 months building social networks instead
The app quietly grew to $500K in revenue with no employees, releases, or support, and 2,000 unanswered tickets
Once they refocused, growth was so vertical the previous vertical line looked horizontal by comparison

“took two days to build we put it out there and it immediately took off right it was absolutely shocking because I buil it on…”
Gaurav Misra · 1:15:00

“I went back to a personal account just opened it and I saw that there was like $500,000 in there and I looked at a…”
Gaurav Misra · 1:16:30

#captions#product-market-fit#founder-story#startups

Takeaway· 1

Takeaway18:00

When Users Complain, That's Product-Market Fit

Shipping a stripped-down MVP surfaces exactly what users care about most, because they complain loudest about it. You can ship fixes for those complaints the very next week, which reads as a hyper-responsive team. Complaints mean people care enough to complain.

A cut-down MVP gets users to tell you what bothers them most
Those complaints become next week's shipping list, making the team look incredibly responsive
People complaining about something is a sign of product-market fit
If nobody complains, that's the worse signal

“next week you can ship in a single week all the things that the user's complaining about right and they're like wow this team is…”
Gaurav Misra · 18:30

“when people are complaining about the thing that means they actually care enough to complain and that's a really good sign”
Lenny · 19:00

#product-market-fit#mvp#user-feedback#shipping

Key ideas

Frameworks in this episode

Cut Scope, Not Quality

Documentation vs Storytelling Safety Framework

Internal Virality for Alignment

Strategic Technical Debt as Leverage

The Public and Secret Roadmap

The Weekly Marketable Feature

People & resources mentioned

People · 6

Resources · 33

Insights & moments

◆Hot Take· 7

Why Right Now Is a Once-in-a-Career Window to Start a Company

A Playground People Leave Isn't a Business

A Startup's Job Is to Take On Technical Debt

PMs Should Own the Product All the Way to Marketing

AI Video Could Return Us to the 1800s

Why He Deliberately Decided Not to Read Books

The Easiest Way to Be the Best Is to Be First

✶Explainer· 3

The Secret Roadmap Competitors Can't Copy

How Snap Ran With Designer-PMs and Almost No PMs

The Documentation vs Storytelling Line for Safe AI Video

❝Story· 4

The Eye Contact Feature That Nvidia Didn't Get

Why Snapchat Declared Itself a Camera Company

Creating Alignment by Making Ideas Go Viral Inside the Company

Captions Made $500K On Its Own While They Ignored It

▲Takeaway· 1

When Users Complain, That's Product-Market Fit

Hot Take· 7

Explainer· 3

Story· 4

Takeaway· 1