Dad to four kiddos, Builder, @XMTP_ co-founder
Joined July 2006
- Tweets41K
- Following1.7K
- Followers30.9K
- Likes29.3K
Whoever did the naming for Claude Duet deserves high praise and a raise.
Sonnet, Opus …Duet
Duet → Do it
Duet → user and agent as a pair
Bravo 👏🏻
@NotionHQ some feedback re: Notion + agents:
1. Right now native page mentions (with @) render as bare Notion URLs to agents using the MCP.
That might lead to them missing critical context entirely, or run extra tool calls just to read page names, which may or may not be relevant to their task.
Ideally page mentions could render as markdown links to agents (or similar) so they can at least get the page names inline without the extra tool calls.
Also adding a few other things below in a thread 👇🏼. Would love to hear what you think @zachtratar @ryannystrom @sarahmsachs @ivanhzhao @akothari @nbaschez
6. Externally connected agents have no Notion-native identifier in the system for use in People properties, comments, etc.
With named personal agents, like Dots @ChatGPT, my @bot(s), or @muse, bringing them in as first-class actors would be huge on many levels. Especially now that MCP events exist…where perhaps an @ mention of them could trigger a routine on their end.
Right now my best workaround has been to set up a separate agents database where they can be named, and @ mention those records. An external app watches for those mentions, and turns them into triggers for those same agents. But it's a lot of extra plumbing that doesn't easily scale, especially since those mentions can't be used in any People-type property.
I really think Notion has got a lot going for it as a "system of record" for personal and company agents, and it's awesome to see all of the attention and effort going into making that a reality. The above are just a handful of things that could make the agent experience even better, so I felt compelled to write them up and send them on over.
I've still got some more things to send your way, but will leave it there for now. As more big ones pop up, I'll just tack on another reply to this thread.
Matt Galligan retweeted
In June, @SamAsante and I launched Typeahead, the first product from our new company.
Today we're releasing our next Mac app. It's called Beam, and we're live on Product Hunt.
I've got Macs all over my house doing different work. Some run agents and local models. Others handle builds. When I want to use one app on another Mac, I usually end up navigating a whole second desktop through Screen Sharing.
Beam brings that app's window onto the Mac I'm using. It gets its own Dock icon and shows up in ⌘Tab. I can type into it and resize it while everything keeps running on the other Mac.
Sam has put a lot of care into making this feel natural. The video shows it better than I can explain it.
Live on Product Hunt → producthunt.com/products/bea…
A step in the right direction for sure. I'm hoping they add support for @NotionHQ custom agents to use ChatGPT subs as that would easily 10x its usefulness to me.
Dots in @ChatGPT can spin up and manage Codex threads to delegate tasks (on-device or cloud), and it's nice! My dot prefixes their threads with ⠕ so I can tell theirs apart from mine. (It's the braile "o").
You can try it out too, prompt shared below 👇🏻
Picking only 9 games from my childhood is basically an impossible task. So here's my pre-high school collection.
Not sure what's been happening w/ my @ChatGPT / Codex but it's burning tokens at an insane rate. I've got a Pro 20x account and yet this morning I went from 40% left to 1% in 90 minutes. Spent one of my banked resets and went from 100% to 30% in less than an hour. This, despite the app saying I've only used 110M tokens today.
1. I've usually got a few threads going at any time, which I've done for a long time.
2. I'm running some sessions w/ Astra Low with a bit of Astra Medium, then 5.6 Sol otherwise
3. No Blender, Computer Use, etc. at all. Just coding & docs work, and light browsing.
4. Not switching between models, or using my account in any other harnesses.
I've halted all processes and feeling stuck…don't even know where to start with how to get this under control and I've already burnt an entire reset w/ this.
Please help @thsottiaux @reach_vb @Dimillian
@bot It's been fun/different/surprising/fascinating using Bot so far. I've built out my squad, and found held up super well thus far. Aaaand now I'm at 6% weekly usage with 4 days left. 😑
I swear before the latest update though, there was a "1 free reset" that I could use, but never pulled the trigger on. And now I can't seem to find it anymore. Am I hallucinating that the reset existed?
cc @benjitaylor @poteto
Ok, so I was only half-crazy. Turns out there is a reset available, but it's on the Grok app, not Grok @bot. Wish I could use the reset for bot!
Got the X MCP server up and running for my agents to use, then set up a @ChatGPT MCP tunnel so I can access it via the web too on ChatGPT.com and it's working smoothly!
Need to get a writeup going on how to set this up next.
Safe to say I've tried my fair share of grid button input device gadgets over the years. Some not pictured here. But enjoying the @work_louder x @OpenAI Codex Micro a fair bit already. Strikes a good balance between simplicity and capability. Plus it's nice to look at.
Just noticed a nice little UX touch in the Claude Code desktop app. Inside a response, if there's a shell command being displayed it includes a little ▶️ play button to run a command directly. Handy!
Monday was my 20 year "Twitterversary" …the same day the X algorithm got updated to prioritize mutuals more.
Feels like a little bit of that old school Twitter magic has come back right in time for my two-decade milestone. 🎉
Rest in peace, friend. @om you were one of a kind.
om.co/2026/06/24/1966-2026/
There seems to be a weird interaction between the installed Codex CLI and macOS apps when it comes to available tools. Even w/ both at latest stable.
At first it was the "automation_update" tool, now the new thread tools.
I've tried every combo of latest CLI using npm and standalone installers, as well as the one that ships w/ the macOS app. I just can't seem to get to a state where Codex agents can use the new thread tools, as was the case w/ automation tools the other day (though that's now working).
Which got me wondering: is there a "blessed" or "best" setup when installing and actively using the Codex CLI and macOS apps, so that all the latest features work in both as described?
cc @OpenAIDevs @romainhuet
