# Changelog #16: Fix errors with AI SRE
You shouldn't have to fix every error yourself. Hand any error to AI SRE and get back a pull request with the fix, from your desk or your phone.
Here's how to fix errors with AI SRE.
## Turn an error into a pull request
**See a new error?** Hand it to AI SRE. It analyses the stack trace, related logs, environment, and recent releases to find the root cause. Connect your GitHub or GitLab and get a pull request with the fix. Review, merge, done.
Hand your next error to AI SRE.
## Prefer to drive the fix locally?
**Use our MCP server or copy a ready-to-use prompt to investigate any error.** Better Stack MCP reads errors, stack traces, logs, releases, and all other data AI SRE can. Our pre-made prompts contain all the important details out-of-the-box and they are customizable. Once done with analysis, just ask your local agent to fix the error.
Install Better Stack MCP server or use our AI prompts.
## Fix errors on the go
**Got a notification for a critical error?** Start a new AI SRE chat from the Better Stack iOS and Android apps. Ask it to list recent errors, pick one, then tell it to investigate and ship a fix as a pull request.
Need to check back on AI SRE? Jump back into the same chat in the app or on Better Stack.
Download our iOS and Android apps.
## Not using Better Stack Error tracking yet?
**Start sending errors to Better Stack in 5 minutes.** Simply swap a token in your Sentry SDK and you're done. No re-integration needed. Each error costs a sixth of Sentry's price.
Start using Sentry-compatible error tracking.
---
## Fresh community content
### Observe Claude Code with OpenTelemetry
Claude Code exports its own usage data over OpenTelemetry: sessions, tokens, cost, prompts, tool calls, and errors. Point the export at Better Stack and see what your team’s AI actually does, what it costs, and where it fails.
### What’s New in OpenCV 5
OpenCV’s first major release since 2018 ships a new DNN engine, real-time YOLO object detection, diffusion-based inpainting, and image captioning with PaliGemma. Computer vision’s standard toolkit just caught up with the AI era.
### [This Tiny Engine Runs Impossibly Big AI Models Locally! (colibrì)](https://www.youtube.com/watch?v=3OqZHYNKc9A";_blank")
Can a 744-billion-parameter model run on consumer hardware? Colibrì is a tiny pure-C engine that streams GLM-5.2 straight off your SSD, tested on a MacBook and an RTX 5090. What’s holding local AI back isn’t your GPU.
### [Apple Just Built WSL for the Mac (Container Machines)](https://www.youtube.com/watch?v=xXtsuDnapqc";_blank")
Lightweight, persistent Linux environments on your Mac, built on Apple’s Container project and optimized for Apple Silicon. WSL-style dev workflows finally arrive on macOS.