AI made writing code faster. It did not make shipping code easier. The bottleneck moved. Most teams haven't noticed yet.
Everyone got faster at generating code. Almost no one got faster at delivering it. Generating code is not the same thing as delivering software.
In 2019 the problem was: Git is hard. In 2026 the problem is: Git is easy. Keeping all the AI-generated code from breaking everything is not. Big changes are coming next week.
@gitlens 18 is LIVE!
If your repo is starting to feel chaotic from coding agents, parallel work, and worktrees everywhere…
GitLens 18 is probably the cleanest workflow upgrade you can make right now.
Better Git workflows for the agent era.
Read the full GitLens 18 release notes here:
https://t.co/eSG0QO5GMv
Your repo after a few weeks of coding agents:
🌿 branches everywhere
🤖 agents everywhere
📂 worktrees everywhere
😵 context nowhere
GitLens 18 is our answer to that.
The Commit Graph became a workbench for parallel development. 🧵
Hot take from Google's DORA research: AI mandates aren't the problem.
Unclear ones are.
When orgs have a clear, communicated stance on AI adoption, individual effectiveness goes up, delivery performance goes up, and friction goes down.
The clarity is the unlock.
Full panel with Google, GitClear + GitKraken: https://t.co/RLznblBOes
Hot take: the most AI-proof developer skill is not ‘prompt engineering.’
It'a making your work easy for other humans to understand. Clean branches. Useful commits. Reviewable PRs. A repo history that tells the truth. That's leverage.
Shameless plug - @GitKraken FTW
My fav is how easy it makes managing & working my submodules (...and avoid those notorious pitfalls).
I still mainly use CLI, but there's something I love about the high-level view it gives that keeps it in my toolbox.
https://t.co/jEkXAiOXrz
We're on the Solution Stage at #LeadDevLondon June 2-3, demoing our AI Agent Leaderboard.
Which agents are actually moving the needle on your team? We'll show you how to find out.