LLenny's Podcast
← All episodes
Nabeel S. Qureshi (founder, writer, ex-Palantir)11 May 2025

How Palantir built the ultimate founder factory

7Frameworks
15Insights

Frameworks in this episode

Insights & moments

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

Hot Take· 4

Hot Take10:00

The Best Companies Send a 'Bat Signal' That Turns People Off

Nabeel argues that great talent magnets put out a distinctive 'bat signal' that intentionally repels some people — that repulsion is the point. He compares it to how OpenAI and Anthropic attract people who are almost messianic about AGI, while Palantir's version was defense and 'save the shire' patriotism during an era dominated by social and mobile apps. Lenny connects it to defining Airbnb's values by being clear about who the company is not for.

  • A good recruiting signal has to turn some people off by design
  • OpenAI/Anthropic attract people who are messianic about artificial super intelligence
  • Palantir's signal was defense and 'save the shire' when hot companies were Facebook and Pinterest
  • Defining company values means clarifying who it is NOT for

they put out this kind of distinctive bat signal and it has to turn some people off. That's kind of the key of a good…

Nabeel · 10:30

it's really important to clarify who this is not for

Lenny · 12:00
#hiring#culture#mission#recruiting
Hot Take57:30

Palantir's Biggest Competitor Isn't Who You Think

Nabeel makes a contrarian point: Palantir's real competitor isn't another vendor but a company deciding to roll its own data infrastructure — a CIO building on a hyperscaler and doing analytics in-house. He also surfaces the human politics: data gatekeepers whose job security depends on being the only person who understands, say, the sales-calculation SQL pipeline, and who resist a platform that makes data point-and-click for everyone.

  • Palantir's biggest competitor is a company rolling its own solution, not a named vendor
  • That looks like a CIO building on top of a hyperscaler and owning analytics internally
  • Some gatekeepers resist because their value depends on being the sole owner of the data pipeline
  • Palantir's disruptive pitch: all data ingested into one platform, easy for everyone — which threatens those gatekeepers

Palant's biggest competitor is a company rolling its own solution right and so the biggest difference would just be a c a c a CIO…

Nabeel · 58:00

let's say I'm the guy who I'm the only guy who understands the the way the sales calculation pipeline works

Nabeel · 57:30
#competition#data platform#org politics#palantir
Hot Take1:10:30

Nabeel's Case Against Disengaging From Defense Work

Addressing the moral objections to Palantir, Nabeel argues that reflexive disengagement is usually the wrong answer. He points to real upside (COVID response, Operation Warp Speed, cancer research) and contends working on defense isn't inherently evil — being 'in the room' can make things like targeting more accurate and reduce error. His broader charge is that tech long refused to look at its own work through a political lens while optimizing, say, short-form video for engagement.

  • Reflexive disengagement is usually not the correct answer
  • Genuine upside existed: COVID response, Operation Warp Speed, NIH cancer research
  • Being in the room can make defense workflows more targeted and accurate, reducing error
  • Tech avoided viewing its own products (e.g. engagement-optimized short video) through a political/moral lens

it's not like working on defense is inherently evil. It's actually a pretty interesting thing.

Nabeel · 1:11:30

again, disengagement isn't the answer.

Nabeel · 1:15:30
#ethics#defense#palantir#tech and politics
Hot Take1:24:00

Contrarian Corner: Going to College Is Great

In Contrarian Corner, Nabeel's belief that most in tech reject: going to college is great, and the 'drop out at 18' advice is mostly wrong. He argues that maybe only ~5% of people (who'd have been Thiel Fellows anyway) should skip it. College is one of the few windows to make deep friendships and spend years reading, writing, and exploring intellectual topics — time that's very hard to reclaim after 21.

  • The 'just drop out and start working' advice is mostly wrong; it fits maybe 5% of people
  • College uniquely enables deep friendships and years of unstructured intellectual exploration
  • That kind of time is very hard to find after 21 once rent and work take over
  • He thinks the ROI is great despite high US fees

I think going to college is great. I think this is a somewhat contrarian view within tech.

Nabeel · 1:24:00

don't drop out of college unless you have a really good reason.

Nabeel · 1:25:30
#college#contrarian#education#career

Explainer· 7

Explainer06:30

What Palantir Actually Does (Finally Explained Simply)

Nabeel answers the question every Palantir employee gets: what does the company do? It achieves tactical outcomes for customers through what he calls the world's best data platform, sold in two flavors: Gotham for intelligence/defense and Foundry for commercial use. It typically sells to very large customers like Fortune 50 companies and governments.

  • Palantir achieves outcomes for customers, delivered through a data platform
  • Gotham is optimized for intelligence and defense; Foundry for commercial use cases
  • Customers are typically very large: Fortune 50 firms and governments worldwide

Um so Palunteer is I the way I describe it right is they um achieve outcomes for their customers very tactically.

Nabeel · 06:30

So they sell a data platform. They typically work with very large customers is the other thing.

Nabeel · 07:00
#palantir#data platform#foundry#gotham#enterprise
Explainer08:30

The Three Traits Palantir Screened Hard For

Nabeel describes the kind of people early Palantir attracted and deliberately screened for. Three traits stood out: fiercely independent-minded people who pushed back and thought for themselves, people with broad intellectual interests, and intensely competitive people with a win-at-all-costs mentality. The screening mechanism was unusual too — a founder had to personally interview you, often as a wide-ranging philosophy conversation.

  • Trait one: independent-minded people unafraid to push back and question the frame
  • Trait two: people with broad intellectual interests beyond tech
  • Trait three: intensely competitive people with a win-at-all-costs mentality
  • A founder (Karp, Cohen, or Lonsdale) had to interview you before an offer; interviews were unpreppable deep-dive conversations

one is like very independent-minded people, people who weren't afraid to push back, who, you know, questioned the frame of everything and thought for themselves…

Nabeel · 08:30

with with Stefan it would be you'd be chatting about philosophy for an hour and a half

Nabeel · 09:30
#hiring#culture#palantir#talent
Explainer16:00

Why Palantir Gave Almost Everyone the Same Title

For a long time nearly everyone at Palantir had the same generic title — forward deployed engineer — with titles reserved for the CEO and six directors. Nabeel explains the logic (which Thiel writes about in Zero to One): titles become a mimetic totem people compete for, triggering Goodhart's-law gaming like starting new products just to get promoted. Removing them kept roles fluid and meritocratic, though it pushed competition toward gaining favor with key executives instead.

  • Titles become something people compete for, producing unproductive conflict and metric-gaming
  • Thiel writes about this in Zero to One; example given of promotion-driven new products at big companies
  • Only the CEO and six directors had titles; everyone else was 'forward deployed engineer'
  • Downside: competition shifts to who can get into an executive's inner circle

as soon as you have these titles, you have a thing that people are competing for and then you get these very unproductive conflicts.

Nabeel · 16:30

everyone's just going to have the same slightly meaningless title which is for deployed engineer and the only people who did have titles were the…

Nabeel · 17:00
#culture#titles#org design#palantir
Explainer19:00

What a Forward Deployed Engineer Actually Is

Nabeel explains Palantir's signature role. Unlike core-product engineers who stay in the Palo Alto or New York office, forward deployed engineers were sent into the field — spending Monday to Thursday physically inside the customer's building, getting their own desk and login access, and building product side by side with the customer's staff. There were two sub-types: classic software engineers and less-code, more 'human-savvy' engineers who could reason about data and navigate executive social dynamics.

  • FDEs literally get a desk inside the customer's building four days a week
  • The role sits inside BD (business development), distinct from PD (product development)
  • Two types: technical software engineers, and 'technical-adjacent' people strong at reasoning about data and reading a room
  • It's radically more embedded than the usual model of occasional customer interviews and Zoom calls

You would literally get a desk there and so that engineer became known as a forward deployed engineer.

Nabeel · 19:30

all of that was given the title of for deployed engineer and it's just an engineer who works with customers

Nabeel · 21:00
#forward deployed engineer#product#sales engineering#palantir
Explainer26:00

How Palantir Turned 'Sparkling Accenture' Into an 80%-Margin Product

For years people dismissed Palantir as a consulting business dressed up as a product company. Nabeel explains how it became undeniably a product: 80%+ margins (versus 20–30% for consulting), and the fact you can now sign up with a credit card. The mechanism was that FDEs were their own first customers, building internal tooling to create value — and eventually a mandate that every deployment had to put a customer on those tools, which forged them into Foundry.

  • Palantir's 80%+ margins are the tell that it's a product company, not a 20–30% consultancy
  • FDEs were 'our own first customers,' building internal tooling that later became product
  • Sean Sankar mandated that every deployment get a customer using the internal tools within ~3 months
  • That painful 3–4 year process of hardening crash-prone internal tools produced Foundry

So they have like 80% plus margins which is not really what you would get if you were actually a consulting company. It would be…

Nabeel · 26:30

so we were our own first customers

Nabeel · 28:00
#business model#product#foundry#margins#services-to-product
Explainer53:30

The Real Secret: Analysis Is Just the Tip of the Data Iceberg

Nabeel argues Palantir's underrated edge was recognizing that inside big organizations, getting and cleaning data is the actual work. Everywhere they went, people waited six to eight weeks just to get data access, and even then the data wasn't queryable. The analysis everyone fixates on is only the top 5–10% of the iceberg; the hidden 90–95% is gaining access, cleaning, joining, and normalizing data — which is exactly where Palantir built product.

  • The big recurring pain point: waiting six to eight weeks just to get data access
  • Analysis is only the last 5–10%; the 90–95% before it is access, cleaning, joining, normalizing
  • Foundry productized each step, e.g. a universal data adapter that reads JDBC, S3 buckets, etc.
  • This white space existed because access was so hard that nobody had solved the problems before

it's this iceberg analogy where the actual analysis is actually just the tip of the iceberg. It's kind of the last five or 10% and…

Nabeel · 54:00

this was the big pain point was we have to wait six to eight weeks just to get data access.

Nabeel · 54:00
#data integration#data platform#foundry#enterprise#insight
Explainer1:05:30

Why Palantir PMs Had to Be Forward Deployed Engineers First

Nabeel explains what made Palantir PMs different: you basically couldn't become a PM without first proving yourself as a forward deployed engineer. The PM for Foundry's ontology, for example, had earlier managed the Airbus factory deployment. The failure mode they were averse to was 'Google Docs syndrome' — writing PRDs and managing rationally from a distance. Successful PMs were best friends with their engineering team and won its trust fast.

  • You could not become a Palantir PM without first proving yourself as an FDE
  • The ontology PM had previously run the Airbus factory deployment
  • They were averse to 'Google Docs syndrome' — managing product via PRDs from a distance
  • PMs succeeded by being trusted best friends with their (often disagreeable) engineering teams

they were extremely careful about only making people PMs who had first proven themselves out as forward deployed engineers. You basically could not become a…

Nabeel · 1:05:30

the ones I saw who succeeded the most were just best friends with their engineering team, right?

Nabeel · 1:07:00
#product management#palantir#hiring#forward deployed engineer

Story· 1

Story32:30

The Airbus Factory Problem That Became Foundry's Ontology

Nabeel tells the story of his Airbus deployment, where the mandate was to help ramp A350 production dramatically (roughly 4x in a year). The team built software to pull Airbus's cryptically-named SAP tables and map them to human concepts — a part, a work order, an aircraft — so any worker could ask 'where is aircraft 79?' and see its station, remaining work, and needed parts. That idea of mapping raw tables to human-legible concepts became Foundry's 'ontology,' still a major differentiator.

  • Airbus's mandate wasn't 'upgrade our data infrastructure' — it was 'help us accelerate A350 production'
  • All data lived in SAP with unreadable table names like S3F1_Z
  • The team mapped tables to human concepts (part, work order, aircraft) into a single legible view
  • This generalized into Foundry's 'ontology,' still a big differentiator today
  • Airbus's CEO said Palantir played a critical part; production ramped ~4x that year

But the overall idea of taking a bunch of tables and then mapping them to human understandable concepts was a very powerful one. And so…

Nabeel · 36:30

the table name would just be like S3 F1_Z or something like that

Nabeel · 35:30
#airbus#ontology#foundry#case study#data mapping

Tool· 2

Tool1:21:30

The AI Tools Nabeel Uses Every Day

In the AI Corner, Nabeel shares the tools he actually uses. Whisper Flow for voice-to-text (great for dictating long LLM prompts, and finally accurate where Dragon and older dictation apps weren't). Claude Code as an addictive terminal-based coding agent that operates directly on the filesystem and executes complex pull requests. And Gemini Pro 2.5, which he'd been testing for a week and calls an excellent model.

  • Whisper Flow: button-press voice transcription, now accurate enough for long prompts
  • Claude Code: terminal-based coding agent that acts on the filesystem and runs complex PRs
  • Gemini Pro 2.5: an excellent model he'd been testing recently
  • He uses LLMs daily for ad hoc jobs, e.g. scripting classification of tax transactions

I love clawed code for developing even though I have my complaints about it.

Nabeel · 1:22:00

So I use Whisper Flow quite a bit.

Nabeel · 1:21:30
#ai tools#whisper flow#claude code#gemini#productivity
Tool1:28:00

Nabeel's Book Recommendations: Impro, the Henriad, High Output Management

In the lightning round, Nabeel recommends three books. Impro by Keith Johnstone — nominally about improvisational theater but really a dense book on creativity, status, and social behavior, which Palantir made every forward deployed engineer read to better read and interact with people. Shakespeare's Henriad (Henry IV, V, VI) as some of the most insightful writing on power and politics. And High Output Management by Andy Grove, valuable for how Grove thinks, not just the tactics.

  • Impro by Keith Johnstone: creativity, status, and social behavior; required reading for FDEs to read people better
  • Shakespeare's Henriad (Henry IV/V/VI): the best books on power and politics, and quick to read
  • High Output Management by Andy Grove: read the full book, not summaries, for how Grove thinks
  • Palantir also used Impro to send to people, per Karp

they made every kind of for deployed engineer read that for the simple reason that I think it just helps you kind of read people…

Nabeel · 1:29:00

I love uh High Output Management by Andy Grove. I just think that's a great business book

Nabeel · 1:31:00
#books#recommendations#impro#high output management

Takeaway· 1

Takeaway59:30

Hire for the 'Extra 20%,' Not the Skills

Reflecting on hiring for his own six-person startup, Nabeel says the hardest thing to find is someone who deeply cares and will go the extra 20% — 'an animal' who pursues the real outcome rather than checking boxes. FANG habits train people to ship code and go home, which is dangerous for a startup's first 20 hires. His answer is mission fit and screening questions like 'what's the hardest you've ever worked to get something done and why' — skills are table stakes; drive is the differentiator.

  • The rare, decisive hire is someone who'll work the evening to get the real outcome, not just check boxes
  • Big-company habits produce people who want a 400k job, set hours, and going home
  • Screen for mission fit and a private reason to care more than average
  • Useful question: 'what's the hardest you've ever worked to get something done and why' — many people have no good answer

somebody who's just checking the boxes and somebody who's kind of an animal in this way, like they'll actually go and pursue and accomplish the…

Nabeel · 1:00:30

you'll ask questions like what's the hardest you've ever worked to get something done and why, right?

Nabeel · 1:02:00
#hiring#startups#mission fit#drive