Shorter Convergence Period
Mark Gurman (tweet, 9To5Mac, MacRumors):
Behind the scenes, though, Ternus has delivered a different message, knowing full well that Apple remains far behind rivals. In recent weeks, Ternus held a closed-door meeting with Craig Federighi, who became Apple’s AI czar following last year’s dismissal of its former AI leader. Desiring more and frustrated by what he sees as a reluctance to embrace AI with the same urgency as competitors, Ternus pushed his software chief to accelerate the company’s road map and move quickly beyond merely making necessary improvements to Siri.
[…]
Apple is already deep into work on next year’s operating systems, Bell (iOS 28 and iPadOS 28) and Poppy (macOS 28), but has changed up its process, I’m told. Apple traditionally separates development of each year’s new OSs into three major milestones known as M1, M2 and M3. Development of each stage runs for eight weeks and is split between four weeks of coding and four weeks of convergence, meaning integrating the features into the operating system.
For this cycle, though, Apple has done things differently for the M1 milestone. Instead of the normal split between two four-week efforts, Apple engineers were tasked with six weeks of coding, exploration and feature development, and then only two weeks of convergence. Employees say they like the change, indicating that it creates more time for coming up with ideas and exploring them before needing to lock things in.
I don’t think Federighi’s problem is that he’s going too slow. Less time for integration sounds like a recipe for even more bugs.
Previously:
- Apple’s Q4 2026 Results
- Siri AI Ships (But Still Beta)
- John Ternus Is CEO
- Glow Leopard
- A Letter to John Ternus
- iOS 27 “Rave” Update to Clean Up Code
- John Giannandrea Leaving Apple
- How Apple Plans to Root Out Bugs
11 Comments RSS · Twitter · Mastodon
"four weeks of coding and four weeks of convergence, meaning integrating the features into the operating system."
Wow, the last time I worked like this, the year started with 19.
> I don’t think Federighi’s problem is that he’s going too slow
The speed is not the problem. The direction is.
Apple patched macOS today to fully remove the compromised eMudhra certificates. 23 days. Shorter convergence is action! /s
Microsoft took 9 days.
@someone +1. Apple's direction is now mostly self-serving and disempowering.
Maybe I'm misunderstanding, but I don't see how this is faster. Instead of 4+4, it's 6+2. The milestones are the same total length.
I also don't think it's necessarily more buggy this way. Are the problems with Apple features typically with the integration, rather than the feature itself? That's not what I've seen.
Personally, I agree with the engineers here. 50% more time to work on projects before needing to merge them sounds like an improvement. Fix the features in isolation, rather than prematurely merging them and then needing to fix them anyway.
(I do disagree with deep integration of "AI", though. It sounds like the Mac is becoming something I just don't want to use.)
Ternus says “go faster”. I assume he doesn’t want software quality to dip even more.
Meanwhile, they’re trying to do the exact opposite of that by cutting AI spending.
@tim They could use the extra development time for polishing/testing, but it sounds like the directive is to use that time to implement more features and then have even less time to test them.
I do think a lot of Apple bugs, certainly the stubborn ones, are with the interaction between different subsystems.
it sounds like the directive is to use that time to implement more features and then have even less time to test them.
One thing I’ve noticed about LLMs at my company is that they will dramatically shorten the amount of time it takes to get a feature/change to a point where it’s ready to be tested by humans, but LLMs don’t dramatically speed up the process of making sure the change is a good one.
They went from a 4+4 to a 6+2 split (for a total of three times). Knowing what I know, a 2+6 split makes more sense than the reverse.
@tim: "(I do disagree with deep integration of "AI", though. It sounds like the Mac is becoming something I just don't want to use.)"
This. For me, it's not *just* AI (much less Siri AI), it's the lack of control of something I've felt I own. Wish I could point at one thing... but when you add the continuous hints, ads, and yes, even (really?) this insistence on not just have an Apple ID but also integrating iCloud.....
To quote a former (now deceased) CEO of Apple, "where is this puck headed"? Hard to say if I'll be there when it gets there.
@Michael": "I do think a lot of Apple bugs, certainly the stubborn ones, are with the interaction between different subsystems."
This too! Key word being "interaction". I may add a word to this... "forced".
> Ternus has delivered a different message, knowing full well that Apple remains far behind rivals.
I got excited for a moment, thinking he was talking about software quality and usability testing.
