Can you convince ChatGPT to send you an iPhone 18 Pro? Let's find out: email your prompt to win@avec.ai before Thursday at 1pm PT!

i just opened @avec desktop for the first time like 5 minutes ago and its already changed my life, so so so so so good

fell out of my chair when i got this email from @avec. also i love their take on the app -> app folder! great taste

In 2026, Afterimage has shipped brand campaigns for @cursor_ai, @turbopuffer, @reve, @every, @pierrecomputer, @avec, @vorihq & more coming soon. This is a lot of work floating around. So we thought we’d put it in one place. Our new site, now live ➡️ afterimage.id

today, i'm excited to share an early preview of the @avec desktop app, which is now rolling out in alpha to folks enrolled in avec's early access program. 5 months ago we released @avec to the app store. we started with a mobile app because it was clear to us that we could build something far better than the status quo there. but as we heard feedback from early users, we realized that there was a much bigger opportunity to do something special on desktop than we initially thought. desktop is not just a bigger screen: it's where you deal with your most important tasks, typically the ones you defer from mobile. what should the interface be when ai can do most of the pre-work for you so you can deal with what remains in a simpler, more focused interface? this demo covers some of the features of avec (e.g. the best prioritization of any email app, stack-based triaging, voice composition), but there are countless more that avec users love and we didn't have time to feature in this demo (e.g. our notifications that are so good you don't need to open the app to know what an urgent email is about, unsubscribe that actually works, or our 1-minute "clean your backlog" flow). one of the things that is most unusual about avec is how simple it is to use, yet how sophisticated it is behind the scenes. we built a 99.99% uptime email processing pipeline that pushes emails through a suite of llms to enrich them and delivers them to you within 10 seconds without ever losing one. we built a cross-device sync system similar to @linear's. we use a dozen or so different llms, and jev, leveraging robust evals and our taste to figure out which one fits for what application. check out the video (or the in-thread clips)! and in the spirit of lifting the veil, today only, ask your questions on how things work and i'll happily answer them here 👇 0:00 Introducing Avec for desktop 0:34 A quick tour of Avec on iPhone 2:05 Why we built a desktop app 2:45 Triaging emails on desktop 3:20 Core interactions: Done, Todo, Voice 3:57 Swipe through email with your trackpad 4:38 Instant sync between phone and desktop 5:34 Todos and keeping track of replies 6:15 A fast, context-aware AI sidebar 7:58 Automatic deadlines and rescheduling 8:45 Teaching Avec what matters to you 9:30 Review and clear low-importance emails 10:08 Desktop alpha rollout 10:49 How get early access

jev understands me on an emoji level

We’re winding down the Notion Mail inbox across web, desktop, and iOS on September 22. We launched Notion Mail with a belief that your inbox should think like you—more personal to how you work and over time, more capable with AI. As Notion agents have gotten more capable, we’ve seen more users hand off email workflows to them. Today, more than half of Notion Mail users manage emails without ever opening their inbox. So, we’re going all in on using agents to run your inbox. We’re grateful to each and every one of you for building your routine around Notion Mail and trusting us with something that mattered as much as your email. More to come! The Notion Mail Team.

got this packaged into the @avec desktop app, rolling out today impressions from 3 days of use: 1. we could’ve done this kind of inference all along with llms, just more expensive. however, the SPEED itself is the real game changer here. being able to rank 100s of threads in half a second unlocks new experiences that simply weren’t feasible before. it will take time to sort out however; most of the demos we’ve seen in the past week won’t ever ship, and the greatest applications likely still have to be dreamt up 2. jev is smart. not super smart. and sadly, unlike llms which give you some kind of plausible rationale for their decisions via chain of thought, it is very much a black box. this means that optimizing a base prompt requires you to do much much larger scale experiments and labeling (llms help here) than when optimizing an llm prompt 3. this specific application is surprisingly useful, though it will take some time for us to apply the same personalization systems we use elsewhere in avec, for the reasons pointed out in 2 above want to try it? download avec, enroll in our “early access program” in settings, and we’ll be in touch soon