Wise Business is now ~31% of cross-border volume, up from ~29% a year ago. Business pays a lower take rate too: 37bps vs 55bps for Personal. Part of the blended take-rate decline is mix, not just price cuts. $WISE

OakNorth outsourcing international payments to Wise is the kind of $WISE growth I like most: the end customer can use Wise infrastructure without ever choosing the Wise app.

A 572-student RCT found a Duolingo-based vocabulary package improved acquisition, but not retention vs traditional instruction. Narrow study, not a verdict on $DUOL. Still a useful reminder: engagement is not the same thing as learning.

Adobe GenStudio can now let ChatGPT, Claude, Codex and Copilot query ad performance, assemble creative and publish approved ads through MCP. That’s “lose the interface, keep the workflow” in practice. $ADBE

eSIMs inside Wise feels either very smart or very distracting. Bull case: $WISE becomes the operating layer for international life. Bear case: feature creep around a great payments network.

$WISE reports Q2 tomorrow. My checklist: • volume per customer • instant-transfer % • transaction revenue vs customer growth • price cuts vs margin If those move together, the flywheel is still doing its job.

Heathrow is using Adobe to improve how its content shows up in AI search. I like the direction: $ADBE is trying to monetize not just content creation, but the discovery layer that sits on top of it.

Adobe started charging paid users’ AI Assistant actions against credits on Oct. 1. That’s the $ADBE datapoint I care about: agentic AI is moving from product demo to usage-based monetization.

Time

When freight rates triple, a 30% NAV discount deserves a second look—but duration is everything.

Debt sets the pressure; inflation and policy credibility decide when the bond market reacts.

Wise became the first non-bank to secure the five licences needed for local operations in Thailand, then launched the account + card there. That’s the $WISE regulatory moat in practice: lots of invisible plumbing before the simple product launch.

For all the talk about AI and Chess, one of the more important $DUOL changes is boring: its nine most popular language courses now reach B2. If Duolingo wants to become a real learning platform, course depth matters more than feature count.

One detail in the Topaz deal I think matters: its Neurostream tech can run complex AI models on-device, not just in the cloud. For $ADBE, better local inference could mean lower friction, lower compute dependence and more AI inside existing workflows.

Premiere on Android is free, with no login and no watermark. That’s a pretty different funnel for $ADBE: remove friction at the edge, then monetize generative credits and storage later. I like the direction.

SYSTRA + $DSY is more interesting than the press release sounds. Virtual twins moving from products into rail and infrastructure widens the lifecycle opportunity. Now I want to see whether these projects turn into recurring 3DEXPERIENCE revenue.

$WISE’s £405M buyback sounds bigger than the pure capital return. Roughly 40% of the shares repurchased are earmarked for employee equity awards. Still useful, but I care more about net share-count reduction than the headline buyback number.

The part of $DSY I’m least comfortable with isn’t the product. It’s capital allocation. ArisGlobal costs ~$1.8B upfront for ~$175M of 2026 revenue. A deep moat doesn’t make >10x sales cheap. This deal needs to earn its keep.