More Code with Jason episodes

298 - AI-Assisted Rails Upgrades with Ernesto Tagwerker thumbnail

298 - AI-Assisted Rails Upgrades with Ernesto Tagwerker

Published 1 Jan 2026

Duration: 2799

AI can aid software developers with tasks like code analysis and error detection, but its limitations require it to be used as a supportive tool, not a replacement for human expertise.

Episode Description

In this episode I talk with Ernesto Tagwerker about using AI for Rails upgrades, AI as an unblocking tool rather than just a speeder-upper, and the da...

Overview

The podcast explores the role of AI in software development, particularly in tasks such as upgrading Rails applications using tools like Railsbump and RuboCops. AI can aid in code analysis, error prediction, and performance improvements, but it struggles with providing accurate, project-specific upgrade plans and estimating the time or complexity involved. While AI can help with problem-solving, mental clarity, and productivity by acting as a "mental lubricant," it should not replace human expertise and needs to be used alongside clear problem definition and common sense.

Additional topics include the challenges of implementing AI, the importance of proper configuration and integration, and the risks associated with speculative code generation. The discussion also highlights concerns related to security, privacy, and the need for responsible AI deployment in production environments. Overall, the podcast emphasizes that AI is a valuable tool but must be used judiciously and in combination with human judgment.

Recent Episodes of Code with Jason

1 Jul 2026 323 - David Yanacek on 20 Years of Innovation at AWS

Modern software development practices emphasize TDD and spec-driven methods, address AI code generation challenges with tools like Kero, explore advanced testing techniques, service architectures, and DevOps strategies, stressing automation balance, observability, and tool refinement for reliability.

16 Jun 2026 322 - Joe Masilotti

AI transforms developer and consulting roles by commoditizing coding, pushing focus toward strategic guidance and app optimization, while challenging solo founders and consultants to adapt beyond code delivery, highlighting AI's strengths in execution over design and the growing need for human-centric, opinionated content and personal branding.

3 Jun 2026 321 - Uncle Bob Martin

Explores software modeling complexities, AI's reliance on statistical prediction versus explanatory knowledge, the balance of elegance and accuracy, abstraction's role in managing entropy and disorder, and the evolution of programming principles like test-driven development and object-oriented design.

More Code with Jason episodes