Ep 147: Scott Wu & Russell Kaplan on the New Era of Software Abundance
Joe Lonsdale: American Optimist Podcast Recap
Published:
Duration: 53 min
Guests: Scott Wu, Russell Kaplan
Summary
Scott Wu and Russell Kaplan discuss the transformative potential of AI in software engineering, highlighting Cognition's autonomous agent, Devin, which drastically enhances productivity. They explore how AI is reshaping industry standards, enabling faster iteration and innovation.
What Happened
Scott Wu, a decorated programmer and CEO of Cognition, describes the company's AI agent, Devin, launched in 2024. Devin autonomously manages tasks that previously required substantial human input, leading to a 6 to 12-fold increase in productivity. By 2026, Devin's usage had already surpassed its previous year's total within just two months.
Russell Kaplan, co-founder of Cognition, traces his journey from Tesla's machine learning team to launching Cognition. He highlights the average age of Cognition's engineering team is 25, with some as young as 17, many of whom are global medalists in programming. The team's youth and talent are pivotal in driving the company's technological advancements.
Cognition focuses on software abundance, which allows turning ideas into reality with unprecedented speed. The team no longer writes code manually but uses English instructions to guide AI agents like Devin. This shift occurred rapidly, over the past six months, and is expected to lead to a hyper-deflationary cycle in software development.
Cognition collaborates with large enterprises such as Citibank and Santander to transform their engineering practices. Devin assists in managing extensive codebases and reducing technical debt, which is crucial for large-scale refactoring projects. The company's approach emphasizes agency and reason, encouraging engineers to focus on creative problem-solving.
AI is revolutionizing the software industry by enabling faster iteration and experimentation, a mindset traditionally associated with machine learning research. Engineers are now embracing non-determinism, allowing for multiple simultaneous approaches to problem-solving. Cognition's software improvements translate into tangible real-world benefits for its partners.
Cognition's government-focused initiative aims to modernize public sector software, addressing inefficiencies such as the prolonged time required for permit processing. The company has secured dozens of FedRAMP deployments with agencies like the US Army and NASA, aiming to streamline and enhance public sector technology.
The integration of AI, particularly Devin, in existing systems without autonomous deployment capabilities ensures that AI-generated code undergoes rigorous review and quality assurance processes. This careful integration helps mitigate concerns about unwanted changes in code and potential risks associated with AI autonomy.
Cognition emphasizes hiring individuals with a high degree of agency, often former founders, to drive innovation and efficiency. The company's special projects team, consisting entirely of former founders, focuses on tackling complex engineering challenges and developing customer-centric products.
Key Insights
- Cognition's AI agent, Devin, launched in March 2024, has demonstrated a significant productivity boost, with one hour of human management equating to 6 to 12 hours of work. By 2026, Devin's usage had already surpassed the previous year's total in just two months.
- The average age of Cognition's engineering team is about 25, and includes many global gold medalists in programming. This youthful and highly skilled team is a driving force behind the company's rapid advancements in AI technology.
- Cognition's approach to software abundance involves shifting from manual coding to using English instructions for AI agents, allowing for faster iteration and creativity. This shift has occurred over the last three to six months and promises a hyper-deflationary cycle in software development.
- Cognition's strategic partnerships with large organizations and government agencies aim to modernize software infrastructure, driving structural changes and improving efficiencies. The company focuses on integrating AI into existing systems to enhance productivity while ensuring rigorous code review and quality assurance.