Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the practical realities of AI software engineering agents in this 32-minute conference talk that cuts through the hype to examine what actually works in modern development workflows. Learn why autocomplete tools like Copilot have achieved mainstream adoption while autonomous agents like Devin and OpenHands remain controversial, with some engineers reporting 10x productivity gains and others dismissing them as sources of noise and technical debt. Discover the key difference between enthusiasts and skeptics: realistic expectations and practical knowledge of effective implementation strategies. Examine which tasks are currently appropriate for today's software agents, identify emerging capabilities expected in 2025, and understand which development activities should remain exclusively human-driven regardless of future AI advances. Gain insights into setting reasonable expectations for AI agent capabilities and develop both practical and intuitive knowledge for leveraging these tools effectively in your software development process.
Syllabus
AI Software Engineering Agents: What Works and What Doesn’t by Robert Brennan
Taught by
Open Data Science