Uncapped with Jack Altman

The Future of AI Software Security | Ep. 39

Daniele Perito is Co-founder and Executive Chairman of depthfirst, an AI-native security platform that understands your code, business logic, and infrastructure to find real vulnerabilities, slash false positives, and give developers actionable fixes in their workflow. Daniele is also Co-founder and Board Member of Faire, where he previously served as Chief Data Officer and helped build the company’s data, risk, and analytics foundations from the early days to a multi-billion dollar valuation. Before co-founding Faire, Daniele worked at Square and was on the founding team of Cash App, where he focused on security, fraud, and risk systems supporting products used by millions of merchants and consumers. We covered: - Inception stories from Faire and Cash App - The ultimate truth seeking machine - Building superhuman attackers with AI - Who wins over time: attackers vs defenders - Why security feels like its own world Timestamps: (0:00) Intro (0:40) The founding Faire insight (4:34) Operational rigor of marketplace businesses (10:39) Starting a company now vs in 2017 (12:01) The inception story of Cash App (16:22) depthfirst’s mission (18:08) AI security landscape (26:10) Security is a fantasy world (31:15) Building superhuman attackers for defense (38:27) Roles of humans and AI in security (39:14) Platform vs pipeline businesses More on Daniele: https://depthfirst.com/ https://www.linkedin.com/in/danieleperito/ More on Jack: https://www.altcap.com/ https://x.com/jaltma https://linktr.ee/uncappedpod Email: friends@uncappedpod.com

DPDaniele PeritoguestJAJack Altmanhost

Jan 14, 2026·45m·Watch on YouTube ↗

📖 CHAPTERS

1. 1

   AI “bears” and why software security is about to get harder

   Daniele Perito frames the coming shift in security: attackers get “abundant intelligence” via AI, turning today’s sporadic threats into scalable, automated ones. The episode’s central premise is introduced—defending software in a world of many AI-powered attackers.

2. 2

   The founding insight behind Faire: take discovery and payment risk off retailers

   Daniele recounts Faire’s contrarian origin: betting on brick-and-mortar retail growth by fixing wholesale discovery and purchasing friction. The key idea was enabling retailers to order with net terms and easy returns, with Faire underwriting the risk using technology.

3. 3

   Finding product–market fit: experiments, then “try before you buy” clicked

   The team explored several approaches (including consignment and rewards) before landing on messaging that instantly resonated. A fast iteration loop—trade show feedback plus overnight code changes—made the winning concept obvious.

4. 4

   Operating a marketplace: rigorous truth-seeking in a chaotic, recursive system

   Daniele explains why marketplace businesses demand unusually high operational rigor: small tweaks can ripple across supply, demand, risk, and onboarding. Data analysis must be paired with intuition to avoid getting trapped in incremental optimization.

5. 5

   Epistemic humility: why A/B tests reveal the limits of “knowing”

   Marketplace iteration teaches a specific kind of humility: strong beliefs routinely fail due to second- and third-order effects. Daniele describes the market as a “truth-seeking machine” for practical product questions, rewarding experimentation over certainty.

6. 6

   Convincing the world: PMF isn’t instant understanding of TAM or inevitability

   Even after early traction, Faire still faced years of explaining the opportunity to investors and recruits. The total addressable market was hard to size, and broad external validation lagged internal conviction.

7. 7

   Starting companies in the AI era vs 2017: faster shifts, higher stakes

   Daniele contrasts the steadier assumptions of 2017 with today’s rapidly changing AI landscape. The pace of change affects strategy, competition, and the psychological intensity founders feel.

8. 8

   Cash App’s early days at Square: outsized impact through security and fraud work

   Daniele shares how he approached Square with a belief that an individual could dramatically change outcomes inside a large organization. He joined the Cash App effort through a security lens, then led anti-fraud improvements that helped the product scale safely.

9. 9

   Why Depthfirst exists: security as a prerequisite for AI safety and control

   Depthfirst is positioned as both a commercial company and a mission-driven effort: without major improvements in software security, broader AI safety goals are compromised. The strategy is a flywheel—secure open source and critical infrastructure while selling enterprise security outcomes.

10. 10

    AI security landscape: toward a unified “AI security engineer” that reasons

    Daniele argues AI changes security tooling from narrow, heuristic scanners to systems that can reason across code and infrastructure. This enables deeper vulnerability discovery, fewer false positives, and consolidation of fragmented security categories.

11. 11

    Attackers vs defenders: imperfect security, lower attack cost, but defenders have context

    Security is framed as economics: perfect defense is impossible, enforcement online is limited, and AI reduces attacker cost—raising attack frequency. Still, defenders can regain leverage by using full-system context and continuous scanning to tilt the balance back.

12. 12

    Building superhuman attackers for defense: Depthfirst’s tech stack and team design

    Depthfirst combines infrastructure-heavy agent scaffolding with deep research (including reinforcement learning) to push beyond shallow vulnerability detection. The team’s composition—infra/security leadership plus advanced AI research—supports both production reliability and frontier capability.

13. 13

    How humans and AI work together + platform vs pipeline lessons from Faire to Depthfirst

    Daniele expects a collaborative model: AI accelerates reviews and reduces the security-vs-productivity tradeoff, while humans retain final contextual judgment. He also contrasts marketplace coordination needs with pipeline-style experimentation, sharing a practical decision habit: study ~30 concrete examples to build intuition fast.