Really enjoyed my conversation with @ColinandSamir about our recent steps into longer-form video. A lot has changed since we talked last, and I think we're both better prepared to start down this path and eyes-wide open about the pros and cons of longer video.

The cost of a fixed level of capability is collapsing — Epoch AI puts it at about 13x per year. And there is no loyalty: usage swung to Astra at launch, then to Opus 5.5 weeks later. No lock-in means new entrants can win on model and price, and incumbents cannot count on anyone staying. My bet: within a year the frontier gap for most use cases stops being perceivable. Then we learn who is paying for performance, and who is paying for the badge.

I was so excited when Emily Bode agreed to design the ring this year. Rings is a small way we look to celebrate some of the most creative people on Instagram. It’s the second year we’ve done this, and it feels good, particularly in the midst of all of this AI mania, to be giving people a small beautiful physical token, in addition to other things, to celebrate their achievements.

Every major platform has had to lean into recommendations — posts from accounts you don't happen to follow — to stay relevant. Going back to a following-only feed would quickly mean a smaller Instagram, with less reach to go around for everyone. That doesn't mean the loss of agency isn't real, or worth solving. Glad to see this becoming an industry-wide direction.

Worth sitting with, including for us. If a standard makes it easy for AI to take context, but hard for the people who built that context to share in the value, they won't share for long. The interesting question isn't open vs. closed. It's whether it's fair in both directions.

Big day for the Muse team with the introduction of Muse Wonder, our most capable model yet. Wonder is built for the long, messy tasks that can take days to finish. It can carry context about you across conversations and keep working toward your goal through blockers rather than giving up halfway. Personally, I'm most excited for us to work out how to use it to help creatives with their work. Technology should empower people, not replace them.

Once iOS27 reaches a majority of our users though I think we'll revisit this, as ideally we wouldn't be (re)building standard components like this.

As is often the case with features like this one, this post inspired a lot of feedback. Turns out the native tab bar now *does* support images. The remaining challenges are: 1. iOS26 recalculates the color each frame, producing a strobing effect as videos play, iOS27 appears to fix this 2. the current tab bar retracts fluidly during scroll to maximize the content viewing area, the native tab bar doesn't support this 3. swipe gestures that start just above the bar trigger unintended tab switches