Ryan Hecht

Here's a video of the Test Track Design Studio experience recreated in Minecraft -- we tried to match the game mechanics as best we could, so if you know how to get a high score in real life you should fare pretty well on here!

@danip717 @unusual_whales The GitHub Copilot CLI team has spent countless hours with Microsoft devs over the past several months ensuring we meet the varied needs of such a large company -- and devs have started preferring us -- More devs at MSFT have used the Copilot CLI than Claude Code for weeks!

@unusual_whales They didn’t cut it for being bad. They cut it because devs preferred it over their own Copilot. That’s not a win for $MSFT

Microsoft, $MSFT, has started canceling Claude Code licenses, per the Verge

@shree_code @unusual_whales Actually, more engineers in Microsoft have been using the GitHub Copilot CLI than Claude Code for a while now! Our team has spent a lot of time with folks across the company to make a tool that exceeds Claude Code on many benchmarks!

@unusual_whales Your own engineers spent 6 months choosing Claude over Copilot, and the fix is to cancel Claude. Tells you who won that comparison.

@shree_code @CaharaDL @ldntownguy @unusual_whales GitHub Copilot CLI (the tool Microsoft engineers are migrating to) has all the latest models! More info here: docs.github.com/en/copilot/ref… (excuse the "Hidden Models," those are staff-only right now!)

@CaharaDL @ldntownguy @unusual_whales Last time I checked, there was news that they had removed the Opus 4.7.

@Insigolon @CodeByNZ Microsoft is transitioning its engineers to using the GitHub Copilot CLI! Our team has spent a lot of time with leaders from across the company the last few months, and internal usage has grown larger than that of Claude Code for weeks: github.com/github/copilot…

@CodeByNZ Don't they get codex for free lol

Microsoft reportedly pulled engineers off Claude Code after the token bills started getting insane. The wild part is AI is starting to look less like a replacement for workers and more like a second workforce to pay for.

@awakecoding @terryaney @_Evan_Boyle @burkeholland Thanks for letting us know -- we'll look into it!

@terryaney @_Evan_Boyle @burkeholland @Ryan_Hecht The exact same environment with Copilot CLI worked with /remote on just 3 days ago. Either something changed on the server or it's a regression in a possible Copilot CLI update that happened since then. It worked before and now it won't let me enable it, with that same error

@_Evan_Boyle @burkeholland Anything going on with /remote command? Screen shot shows: 1. Attempt at /remote on 2. user command results 3. My github account labeled as 'personal'. I've ran remote sessions in this repo before. Not sure why not working or where to look.

@FiBelleFi Have you taken the various cable cars over to Breuil-Cervinia yet?

There’s some fantastic Italian food in Zermatt since Italy is just over the mountains. Comfort food!

@FiBelleFi lol this thought would not leave my mind the entire time I was in Zermatt

Ngl I never realized the extent of the possibilities in a potential EPCOT Switzerland pavilion until I came here.

@CameronFoxly What, you didn't make your own animation software suite for this one?!

In the age of AI slop, some of us are still making things by hand, I promise ❤️ Octocat for the new GitHub Copilot App was modeled, rigged, and animated in Blender...

@DanBerger_ @tomwarren Copilot CLI is a comparable product to Claude Code -- we've spent a lot of time with engineering leaders and IC's across Microsoft to make sure we filled any/all functional gaps they were seeing :) Internal benchmark/evals have us at parity or even scoring slightly higher!

@tomwarren For sure this is a cost cutting decision masked as dogfooding. Copilot and Claude are two very different tools.

scoop: Microsoft is starting to cancel Claude Code licenses. Engineers in Microsoft's Experiences + Devices team will have to transition to GitHub Copilot CLI by the end of June. Details in my Notepad 📒 issue, live now for subscribers 👇 theverge.com/tech/930447/mi…

@pat62016 @FiBelleFi Hence my comment, "Guess there's no hope for Tomorrowland Speedway!"

@Ryan_Hecht @FiBelleFi That was specifically to comply with California.

You're telling me that the thing that made them decide to electrify Autopia was an EMISSIONS SCANDAL?! (Guess there's no hope for Tomorrowland Speedway then 😥)

@FiBelleFi Ugh I love Amsterdam so much

I took a thousand canal photos today ❤️

@FiBelleFi Well, you can't live in an ETF!

Chapek made only a 4% return on the sale of his home which is crazy low considering he owned it for 5 years. Should’ve bought some S&P 500 index ETFs instead Bob….thats up 80% from 2021!

@GHOSTawyeeBOB @KnightedNate 👋 hi, I'd like this crap! ...I'd also like some light rail for intra-metro-area trips. - an Orlando resident who's tired of driving so much

@KnightedNate No one wants this crap.

With Brightline in the news (unfortunately), I figure my mock-up of a 30-year wishlist for commuter/regional rail in Florida would be fun to show off Some of this is already on its way; some will take years. Transit comes in steps. One day, Florida will catch up with the world

@davidfowl I uh, think I've made 3 or 4 such systems this year so far? @patniko and co keep adding new features to the SDK that render parts of my personal system obsolete!

I’m gearing up to build my own agent orchestration system. Are we all doing this now?? What stage of grief is this?

@theo I got into the habit of asking Opus "Have a GPT subagent take a critical look at this" whenever doing anything nontrivial. We ended up adding this as a first-class feature to CopilotCLI: github.blog/ai-and-ml/gith…

It's so hard to describe the vibe difference between Opus 4.7 and GPT 5.5 (for coding) GPT is smarter and can unblock you, but it gets stuck in stupid ways and strangles itself with context sometimes. Opus will go down the most insane paths and refuse to acknowledge obvious answers, but it understands intent better and has more taste. Whenever I use one for more than an hour, I always reach to the other to "clean up". Best part? All of this changes every few weeks 🙃

@T0k3nR1ng @WDWNT I don't think it's false. 51,000 model calls per day would NOT be the largest I've ever seen from non-abusive use!

@WDWNT This is stupid and no way its true. Likely someone had a ton of sub-agents running or there is some other metric that is being misunderstood.

One Disney Employee Used AI Chatbot 51,000 Times a Day wdwnt.news/63t_26

@SteveJabs @WDWNT That being said: counting number of model calls per user isn't really the best metric. I expand on this a bit more here: x.com/i/status/20492…

@SteveJabs @WDWNT 51,000 model calls per day is NOT the most I've ever seen from non-abusive use