Uncapped #39 | Daniele Perito from depthfirst
Uncapped with Jack Altman
12 HOURS AGO
Uncapped #39 | Daniele Perito from depthfirst
Uncapped #39 | Daniele Perito from depthfirst

Uncapped with Jack Altman
12 HOURS AGO
In this episode, Daniele Perito shares insights from his journey co-founding high-impact companies like Faire and Cash App, before diving into the evolving role of AI in cybersecurity. From early startup pivots to building intelligent defense systems, the conversation explores how truth-seeking mechanisms—both in markets and technology—shape long-term success.
Daniele recounts how Faire capitalized on the overlooked potential of brick-and-mortar retail with a 'try before you buy' model that quickly validated product-market fit. He contrasts today’s volatile startup landscape with the more stable environment of 2017, then traces Cash App’s origin from a risky Hack Week experiment to a secure, scalable platform powered by machine learning and proactive fraud detection. This experience informed depthfirst’s mission: using AI to secure open source software by identifying real vulnerabilities with minimal false positives. AI is transforming security, enabling deeper code analysis and context-aware testing through reinforcement learning and AI agents. While attackers gain efficiency from AI, defenders benefit more through superior context and transparency. The future lies in hybrid systems where humans maintain oversight, making faster, confident decisions within secure frameworks—blending operational rigor with innovation.
00:44
00:44
Retailers could order without paying for 60 days and return unwanted items, creating trust and reducing risk.
07:19
07:19
The market is a truth-seeking machine for business-related questions
14:35
14:35
An anxious and paranoid mindset reduced fraud losses by 80% at Cash App
16:22
16:22
depthfirst aims to secure open source software and create positive impact through AI-driven vulnerability detection
18:08
18:08
AI can detect previously undetectable security issues in code
28:40
28:40
AI can verify its own security findings by showing assumptions and process
31:15
31:15
Reinforcement learning combined with large language models can create a superhuman hacker for defense.
41:52
41:52
Analyzing 30 anecdotes provides more insight than 3,000 emotionless data points.