Claude Code's latest bug made me laugh - and then cringe. The AI coding assistant was randomly running git reset --hard origin/main against project repos every 10 minutes, essentially nuking developers' work. Classic AI overconfidence.
This hits close to home as I'm deep into building AI tools, but also watching my best SEs get mesmerized by AI tooling right now. They're spending hours perfecting prompts, building custom workflows, and evangelizing the latest AI demo features for prospects.
Here's the problem: while they're playing with shiny objects, they're missing the fundamentals that actually close deals.
I recently saw one of our top performers spend days building an AI-powered demo that could "dynamically adjust to customer use cases." Impressive? Sure. Effective? They just wanted to see our alerting work with their existing Prometheus setup.
I had to have the conversation: AI tools are supplements, not replacements for SE craft. Understanding customer pain, building trust through technical depth, and guiding complex evaluations - that's still human work.
The irony? The companies buying from us are digital natives scrambling to implement AI everywhere. They need SEs who understand both the promise and the pitfalls, not engineers who've been replaced by their own tools.
Don't let your team mistake productivity theater for productivity. The fundamentals still matter, even when the future feels automated.
What conversations are you having with your SEs about AI boundaries?