Real-time Coding Agent o11y. Now you can see real-time telemetry on cost, tool calls, and the journey of all the AI-code in your repos from prompt out to production.
"We only have the bills" - how a lot of engineering leaders we work with feel about measuring ROI before Git AI.
It's difficult to do this right because of long spanning sessions contributing code to different commits, brainstorming sessions that don't lead to code, stacking..
Uber’s COO has said that it’s getting “harder to justify” its AI costs because there was no way to show a link between AI spend and any meaningful increase in useful features. This is the first time I’ve seen a company say this directly.
businessinsider.com/uber-coo-andre…
1.4 shipped huge improvements to how Git AI connects LoC to Agent Sessions.
- Tool-call level granularity. Connect each AI-line to the tool that generated it
- Support for non-linear sessions (subagents, interruptions, rewinds)
something is coming
usegitai.com/changelog/2026…
Git AI now tracks and attributes changes that result from Agents running shell commands.
super cool to see the community take on something this big and useful 🚀
Retroactively calling this a launch week because v1.2 is full of so much good stuff:
🫡 Maximum compatibility — no more Git hooks or git wrapping. It just works.
🚀 No overhead — attribution calculated off Git hot path
🛸 Support for Background Agents
Exactly what Git AI was built for github.com/git-ai-project…
I think the signal we want in RL extends way further than sessions…
- what human overrides were made
- was the AI code approved
- what got changed in code review
- is the generated code still in prod a month later?
- did AI code cause a bug or incident?
It’s about 100x harder to do track AI code through the SDLC - but that’s where the real reward signal is
Your team uses 4 different coding agents, and you're starting to spend a fortune on them.
Which one produces the best code? Which do devs actually use?
cubic now tracks which lines of code AI wrote, and which model and harness produced them - partnering with Git AI.
Over the last few weeks companies have begun running Git AI in their custom background agents. Turns out you can't build an autonomous software factory without observability and attribution for code and decisions.
Super cool to be listed here by our friends at @ona_hq
Community has been cooking this week 🚀
New contributors added support for @kilocode and @Kimi_Moonshot as well as porting the AI-blame extension to the @jetbrains ecosystem.
Seeing this trend a lot of this across teams that use Git AI: enterprises are investing in their own background agents. I was skeptical at first, but I get it now.
Building and optimizing the factory will be the work of platform teams for the next 5 years.
Agentic software engineering adoption is on fire at @Uber. 1,800 code changes per week are now written entirely by Uber's internal background coding agent, and 95% of our engineers now use AI every month across all the tools we track.
This is a real reset moment for engineering;
@saturnial I mean, I've rarely seen such a detailed PRD that mixed product requirements with technical specification, was kept up to date over time, etc.
you still have some spec, or you can iterate on the feature with the agent directly and the spec lives in prompt history and you can use
Also huge Shot to the @AmpCode team. Their new plugin system takes a very different approach than most Agents, but we think they have the right model.
Excited to build more with this team!
Looks like @AmpCode was here 🚀.
The Git AI <> Amp integration is now live. Track all the code written by Amp — along with the prompt that generated them.
Git AI now lets you use past prompts as context — across agents. Whenever you're building on top of AI-code the original prompts help inform what the agent writes next.
Claude knows and adheres to the requirements you gave Cursor. Codex knows about an architecture decision you made in OpenCode. All your agents remember your intent and why they wrote code the way they did.
Get more of your team on Git AI and all your team's agents get smarter.
4K Followers 3K FollowingThis account only follow anything related to early or alpha projects. Never expect too much or your decision will never right.
2 Followers 175 FollowingGazing at stars. Thinking about dark forests. Chief of Staff at a startup you'll hear about soon. Probably not a large language model in a trench coat.
4 Followers 100 FollowingSystems Architect. I build data pipelines, scenario engines, and revenue infrastructure. Case studies at https://t.co/heXZ4LFAUO