Ex-VP Business Development at GoCardless, Former Palantir Engineer
LENNY'S PODCAST
Palantir's Secret
The 30% Founder Effect
30% of PMs that leave Palantir start a company (vs 18% at #2)
Palantir PMs rank #1 when promoted at their next role
The company is intentionally a founder factory
The culture, structure, and customer engagement model turn engineers into entrepreneurs
Why it mattersPalantir doesn't just hire great people. It systematically trains them to think like founders. Every engineer gets founder-level customer exposure and P&L responsibility.
Culture & Hiring
What Palantir Screened For
Independent-minded: People who questioned everything, pushed back on framing, had strong convictions
Broad intellectual interests: CEO Karp quotes Habermas; they hired thinkers, not just engineers
Intensely competitive: Win-at-all-costs mentality. "We want to solve the hardest problems in the world"
Mission alignment: Attracted talent from military, intelligence, Special Forces — people who understood serving something bigger
The "bad signal" strategy
Palantir's focus on defense and military ("Save the Shire") repelled most of the valley but attracted exactly the right people. The distinctive, off-putting positioning was intentional.
No titles, no promotion races
Everyone generic title. No hierarchy to game. Eliminated Goodhart's Law problems where people optimize for metrics rather than impact. Direct impact on product quality.
"When you define your company values, you need to be clear on who this is NOT for. That's the key to attracting the right talent."
Forward Deployed Engineers
The Secret Weapon That Scales
What they are
Real engineers sent to live at customer sites (Mon-Thu). They understand the customer's actual business, build custom solutions, maintain close relationships. Not sales engineers — true builders with product autonomy.
Why they work
Each FDE serves 3-5 customers simultaneously. As product matures, ratio improves. Early stage: 5-10 engineers per customer. Mature: 1-2 engineers per customer. This is your North Star for product leverage.
The key to getting it right: FDEs must be real engineers who can build new product, not just deploy existing ones
Deep business understanding: Know how your customer actually makes money. Hospitals want to turnover patients like restaurants turnover tables
Personal relationships: Be on texting terms with buyers. See them as humans to help, not customers to extract value from
In-person presence matters: Relationship depth, trust, and problem discovery require physical proximity
AI changes the mathLLM coding tools (Cursor, Windsurf) make FDEs more scalable. One person can now do the work of two. This makes the model more accessible to earlier-stage startups.
Product Playbook
Talking to Customers the Right Way
Listen to what they're asking for: Customers rarely know what they need, but they know their pain
Don't anchor to your idea: Be willing to drop your assumptions and treat the problem as a blank slate
Reframe and test: Retool changed "supercharged Excel" to "build internal tools easily." Same product, different buyers, massive traction
Murder boards: Write a 2-page plan, invite smart people unfamiliar with it, let them tear it apart. Principles must be controversial
The principle testIf everyone agrees with your principle ("move fast"), it's not a principle — it's a platitude. Good principles should make half the room disagree.
Contrarian
Why Palantir's Model Looks "Wrong"
✗Only hire "nice" culture fit employeesINSTEAD →✓ Hire intensely competitive, independent-minded people who question everything. They'll build better products.
✗Maximize titles and promotion paths to motivate peopleINSTEAD →✓ Remove titles entirely. Let impact speak for itself. Stops politics and metric-gaming before it starts.
✗Build a product, then sell it to customersINSTEAD →✓ Have engineers live at customer sites. They'll discover the real problem and build something that actually solves it.
✗Customers tell you what you should buildINSTEAD →✓ Listen to their problem, not their solution. Be willing to reframe completely. They might not know what they actually need.