You've probably already approved one without realizing it. 👀
Agent-generated pull requests pass the tests and show clean diffs, so you merge. That's exactly the problem.
This checklist catches what they hide: gamed CI, security gaps, and bugs that slip past green checks.
https://t.co/8IpI883Hii
API access for enterprise admins to request and download billing usage reports in CSV format is now generally available.
• Programmatically create the same billing reports shown in the UI
https://t.co/Sh9VNweuJd
Everyone's rewriting their GraphQL gateways in Rust and Go for performance.
We stayed on .NET.
Fusion 16 now ranks #2 in federation benchmarks, ahead of two Rust routers and a Go one. Don't be fooled: .NET is fast 🚀
#dotnet#graphql
https://t.co/n5XVEFNHsV
1 million context window and configurable reasoning levels are now available in GitHub Copilot for @code, Copilot CLI, and Copilot app developers.
https://t.co/X4zkWqMse2
GitHub Copilot supports one-million-token context windows and configurable reasoning levels for handling larger codebases and deeper problems.
• Available now in VS Code, Copilot CLI, and the GitHub Copilot app
https://t.co/Gz5Sfk4JBg
Just used 1.3% of my monthly balance on a single GitHub Copilot request to rename some types and that was after I stopped the agent because it was overreaching.
Aspire really lights up your developer experience. Today is a huge milestone as we have GA support for Typescript.
If you haven’t given it a try yet, take 5 mins to play with it!
Super easy to get started:
https://t.co/KikC1uObWN
The outer loop is just as painful for agents as it is for humans. The only difference is that I can go do something else while its running. That doesn't negate the need for improving the underlying systems.
We have some exciting announcements during the Visual Studio session at Build next week. Make sure to tune in online if you can't be there in person. https://t.co/im8BN9QNar
Some serious improvements went into the code review agent on @github. It’s roasting the dev team PRs 😅. It was so bad before now it’s become really valuable.
The speed at which this tech is improving is so dramatic. This is the new normal.
One of the underrated reasons Linear is so popular with so many people is they have an internal target that nothing in their interface should take more than 300ms to render. They keep fixing regressions whenever it happens.
It’s very hard to retrofit this culture: look at JIRA…
Situation 1: dev A thinks approach X is correct, dev B thinks Y is the right way. They argue and try to convince each other.
Situation 2: dev A thinks approach X is correct, tells the LLM to implement it.
There is SO MUCH learning in Situation 1, lost when using LLMs....
I find myself doing a lot better work, being more satisfied, and also learn a lot more+faster when I do *the hard work* and don’t outsource it to AI.
As in, I’ll use AI as a *tool* with substasks, additional research: but I don’t turn off my brain or kick back, assuming it can do the work for me.
Every time I “hand over the” hard work part to AI and mentally turn off, I either regret it or find myself eventually needing to go back and spend more time on it.
I also see slop work coming out from people who assume the AI does better work than they would.
@RhysSullivan *Really* good tests. I've done this a couple of times now and the agent will lie and say it's done. It will even mock enough data to make these tests pass but not be a real implementation.
This is similar to what anthropic saw when writing the C compiler.
The aspire team is beginning to find a stride working with ai to build production ready software (this is challenging). This next milestone will be the first one where we have docs automated, better skills for creating pull requests and doing code reviews, doc writing and testing, end to end testing.
Lastly the team has had some really “fun” learnings about how too much ai with little understanding of what was emitted can cork back to bite you 😬