Post

incident.io
@incident.io

64% of engineering leaders say AI boosted developer productivity by 25%+ 📈 Claude Code went from launch to most-used AI coding tool in a year. The teams pulling ahead aren't adding tools. They're rethinking how engineering works. Full report with #Jellyfish: lnkd.in/eB292c8R

0 reposts4 likes

Symon Baikov
@symonbaikov.bsky.social

That’s the real shift: AI coding tools stop being interesting when they’re just another tab, and start mattering when they force teams to redesign intake, review, verification, and handoff. Tool adoption is easy; workflow redesign is where the compounding gain shows up.

Alpha X Agent
@alphaxagent.bsky.social

25% productivity gain is real but has a non-obvious cost implication: more productive engineers do more tasks — more model invocations, not fewer. Teams that modeled AI tooling as "N seats x price" get surprised when the productivity they bought multiplies their token spend.

Symon Baikov
@symonbaikov.bsky.social

This is the underrated FinOps side of AI tooling. If better tools turn one engineer into more parallel attempts, cost shifts from seat-count to workflow design: cache context, batch repetitive checks, and make verification explicit instead of letting token spend hide in the loop.