Why isn't This AI Thing Working podcast

The Gap Between Vibe Coding and Production-Ready AI | with Anuja Gokhale

0:00
44:24
Spola tillbaka 15 sekunder
Spola framåt 15 sekunder

Everyone has seen the demos: someone vibe codes an app in a day and ships it. So why does AI feel slower and messier inside a real engineering org? HungerRush CTO Anuja Gokhale has a clear diagnosis: vibe coding a prototype and running software in production are two different things, and AI amplifies whatever system you already have. Without clean data, strong test coverage, and automated pipelines, faster code generation just moves the bottleneck somewhere else.

This week Jess Lampe talks with Anuja about her team’s journey bringing AI into the software development life cycle. She explains how AI can function like a junior engineer, why her team spends more time planning before any code is written, and why a human in the loop is a requirement, not an option. She also shares how a cross-functional hackathon produced a feature that went from idea to production in two days, along with a machine learning model built by engineers who had never worked with one before.

They get into the realities: reviewers juggling several AI sessions at once, the risk of getting less diligent as code volume grows, and why maintainability still matters even when AI is reading the code. Anuja closes with a challenge for the industry: if AI writes more of our code, how will today’s junior engineers build the hard-won judgment to own it, including when something breaks in production at 2 a.m.?

Fler avsnitt från "Why isn't This AI Thing Working"