Here’s what our cloud hosts had to say about this week’s episode: 346: Zuckerberg Finally Finds His People, They Are All AI Agents
Justin Brodley
Another week, another GPT model. I appreciate OpenAI’s drive to accelerate innovation, but the pace of point releases makes it genuinely hard to keep up. It’s nearly impossible to tease out the real differences and strengths of each model before the next one drops. Good to see them getting into the Excel space, but I’m holding out for PowerPoint support and the next version of Cortex before I kick the tires on OpenAI again.
Amazon is pushing back on Financial Times reporting that AI-generated code is behind failures on and AWS, pointing the finger squarely at the engineers doing the prompting. I’m not buying it. This is a shared responsibility problem, the same way AWS security is shared between AWS and its cloud customers. Engineers should be shipping small, incremental changes with the help of tools like Kiro or Claude, backed by solid testing processes and a CD pipeline that catches rising failure rates and rolls things back automatically. Higher-impact changes should require more eyes and more PR reviews. This looks like a failure to properly account for AI in the SDLC, and Amazon is almost certainly not alone in that. But pinning it entirely on the engineer is blame culture at its worst.
🎧 Listen to the full episode
Leave a Reply