JJ Englert

I've been using @claudeai Tag for the past 2 weeks. Two things matter. Ambient mode is an actual unlock, and the multiplayer experience is dope af. By default it works how we all know. You ask, the agent answers. Ambient means Claude stops waiting to be tagged. It flags what it thinks you need to know from the channels it's in and the tools it's connected to, and follows up on unresolved tasks on its own. You can also give it standing work. Anthropic calls those routines. You name the schedule, like "every two hours, check the alerting dashboard and post anything new." Or point it at a channel to watch, or a PR to follow. It only posts when something actually changed. That's why Open Claw and Hermes took off earlier this year. Work happening without you asking for it. Think about support email. Today I have skills that answer tickets, but I have to run the skill for anything to happen. With a routine, Claude sweeps the channel on the schedule I set, answers what it can, routes the rest, and tags me on anything it shouldn't decide alone. That's when it starts to actually feel like an AI teammate. I also LOVE Multiplayer mode. And now I want everything to be multiplayer. We setup claude tag into a channel for our marketing website, filled with some of my non-technical teammates that need to control the website (careers, blog, marketing pages, etc) When Brett, our ops lead, asked Claude for new filters on the careers page, Claude got stuck initially. So I jumped into the same thread and walked it through, and Brett watched the whole thing happen (learning). That's new. Technical and non-technical people working one agent together, in one thread, where everyone sees what it did and can correct it. Picture your engineering channel with Claude subscribed to your PRs. CI finishes or a review lands, it posts, and whoever's free grabs it. Your CTO has visibility without asking anyone for a status update. At any point somebody can step in and say do it this way instead. My take If you're really nerdy you could probably build a Slack agent that does this better yourself. Out of the box, Claude Tag's memory and its tools are just really good, and that's what makes it worth reaching for. I think it's the start of a lot more multiplayer agents. One agent with real tools and permissions, with your whole department in the thread with it, is a really cool experience. Full video below: 00:00 What Claude Tag Is 02:52 Setting Up Claude Tag 04:42 Teams And Enterprise Only 05:03 Ambient vs Non-Ambient 05:43 How Ambient Mode Works 06:38 Support Email Use Case 08:32 Multiplayer Mode With Your Team 09:19 Shipping Career Site Filters 13:59 Website Security And Social Monitoring 15:34 Hooks, Security Checks, And CI/CD

There have historically been two career paths for engineers: ic & manager. I'm offering a third path: creator. The job is... - 1/3 ai engineer - 1/3 ai creator - 1/3 ai educator Hot take: being a creator is one of the highest status jobs in a sea of abundance, where earning & nurturing trusted distribution has never been more important. But many of the best creators don't just want to create content. They want to practice what they preach, while using the learnings from their craft as the foundation for the content they create. Think @theo, @trq212, @signulll, @bcherny, @dylan522p and many others. Which is why we've created & hiring for the Applied AI, Engineer & Creator at @tenex_labs. You'll solve complex problems with AI for some of the biggest companies in the world, while leveraging those learnings to create content online & lead AI workshops for F2000 enterprises. Must-haves for the role: - Cracked 10x engineer - Strong communication skills Nice-to-haves: - A student of X - Experience creating content online - A small, but mighty following online And yes, I'm looking for a unicorn. Apply here: jobs.ashbyhq.com/tenexlabs/18fd… Tag anyone below you think could be a good fit.

Two awesome people discussing AI. You should watch it!

@clairevo @tenex_labs @businessbarista lets gggggoooo!

The content machine at @tenex_labs is cooking, and @businessbarista is at the center of it. He did the impossible: figure out how to use AI to draft social posts that don't suck. In this ep, he shows how to: - scan your company data for daily content ideas - use an AI interview panel to get his thoughts - train the model on sounding like him - run a critical editorial council and editing loop - turn his feedback into skills so the system gets better But my favorite part of his flow is decidedly human: the reply-guys slack channel where the whole team hypes up eachother's social posts. Brough to you by our awesome sponsors: 🔥 @firecrawl - Power AI agents with clean web data 🤝 @CustomerIO - Build customer engagement campaigns from a single prompt Full ep on YT: youtube.com/watch?v=1_jluk…

@TylerMorganMe haha no. Just a fun thing to see hanging around our offices.

@JJEnglert Thought you had some legal issues. good hook. You planning to read through the whole thing?

Reading up on the constitution

@TylerMorganMe lol fair

@JJEnglert I had a quick "Why?" moment. lol

@theaaron Hahah

@JJEnglert it explains a lot of the how and why behind how Anthropic operates too, not just their model behavior

@theaaron @OpenAIDevs That’s a really cool idea

i hosted a workshop where my slides were controlled by the @OpenAIDevs GPT Realtime voice model that injected live audience data into the deck...it cost $0.67 to test and run for about an hour

@anitakirkovska @its_avareed He’ll yeah

I think I've finally built @its_avareed to feel like a real AI coworker that has context on everything I do for work, is cost-efficient and proactive 👀👀 She now: 1/ joins all my meetings by herself 2/ assigns tasks to herself from what I say out loud (sometimes it's cringe when I call her in the middle of a meeting) 3/ does follow ups autonomously 4/ captures all my business context end to end from Notion, meetings, Slack 5/ plans my week, my days, and pushes me to do more 6/reports on what changed proactively She does this with literally just skills and hooks, and we're spending between $15-20 a day with open source models on very heavy tasks. That might be high for some, but for my use-cases I think it's pretty solid. Sharing the whole setup this week - lmk if you wanna get it!

@ArmanHezarkhani @cvpayne Such a cool experience!

I spoke with @cvpayne about the Kimi K3 Now, there's another open-weight chinese model. China is not slowing down. We have to let innovators innovate. 1. Stop the doomerism. 2. Build up public sentiment and trust. 3. Companies need to know how to use AI and people need to know its not coming for their jobs!

@rileybrown We’re here, baby!

The TBPN-ification of Fox Business

How cool is this! One of the perks you get for working with incredibly talented people

@cvpayne Amazing stuff! We're always here to help translate complex AI topics into things the every day person can understand. It's what we do for a living!!! Think of us as your AI Geek Squad 😉

Thanks He was fantastic

@surim0n @Microsoft ughhhh that's such a Microsoft thing... So weird. On paper they have everythig you need. Then you get into the details, the error codes, etc and it falls apart, every time.

@JJEnglert @Microsoft Did copilot studio training for an 11-figure company. After making a few simple agents (live), it kept refusing to do the work. “sorry we can’t do that right now, please try again later”. The amount of trust lost is unrecoverable.

Currently going DEEP into the @Microsoft stack (specifically Copilot Studio) for client trainings. Some thoughts.... They're catching up. So much what what has worked for other companies, (Cowork) (Agents) (Automations) they have their own version of. And with control of most of the enterprise users, and their data, it's a really easy sell to enterprises. Their software even just from 2 months ago to where they are today, has jumped significantly. Every day you see a new thing, new feature popup. Almost in a bad way cause it's changing so fast. But these changes, currently feel like they've found their rythem. It's still over complicated at times. Still too many options to pick from, which overwhelms many users. But it generally has everything an enterprise needs to bring their workers into the "doing more with AI" age, and then some. And then the model agnostic approaching, offering a control plane on top of it (even in a small way) feels like the right move at the moment. Who are the best Microsoft builders that you know? Would love to build up more of this network.

@donlangley @businessbarista @cvpayne hell yeah!

@businessbarista Great segment on @cvpayne's show today on Kimi K3 and the market! Keep up the great work!

If you have non-builders building stuff at your company and want a better process for how that works, start here

@LoopOnChain @claudeai I don't think it does, no!

@JJEnglert @claudeai Does ambient mode use significantly more tokens? Thanks for sharing man this is great! Actually just started planning to implement claude tag with a client!

I've been using @claudeai Tag for the past 2 weeks. Two things matter. Ambient mode is an actual unlock, and the multiplayer experience is dope af. By default it works how we all know. You ask, the agent answers. Ambient means Claude stops waiting to be tagged. It flags what it thinks you need to know from the channels it's in and the tools it's connected to, and follows up on unresolved tasks on its own. You can also give it standing work. Anthropic calls those routines. You name the schedule, like "every two hours, check the alerting dashboard and post anything new." Or point it at a channel to watch, or a PR to follow. It only posts when something actually changed. That's why Open Claw and Hermes took off earlier this year. Work happening without you asking for it. Think about support email. Today I have skills that answer tickets, but I have to run the skill for anything to happen. With a routine, Claude sweeps the channel on the schedule I set, answers what it can, routes the rest, and tags me on anything it shouldn't decide alone. That's when it starts to actually feel like an AI teammate. I also LOVE Multiplayer mode. And now I want everything to be multiplayer. We setup claude tag into a channel for our marketing website, filled with some of my non-technical teammates that need to control the website (careers, blog, marketing pages, etc) When Brett, our ops lead, asked Claude for new filters on the careers page, Claude got stuck initially. So I jumped into the same thread and walked it through, and Brett watched the whole thing happen (learning). That's new. Technical and non-technical people working one agent together, in one thread, where everyone sees what it did and can correct it. Picture your engineering channel with Claude subscribed to your PRs. CI finishes or a review lands, it posts, and whoever's free grabs it. Your CTO has visibility without asking anyone for a status update. At any point somebody can step in and say do it this way instead. My take If you're really nerdy you could probably build a Slack agent that does this better yourself. Out of the box, Claude Tag's memory and its tools are just really good, and that's what makes it worth reaching for. I think it's the start of a lot more multiplayer agents. One agent with real tools and permissions, with your whole department in the thread with it, is a really cool experience. Full video below: 00:00 What Claude Tag Is 02:52 Setting Up Claude Tag 04:42 Teams And Enterprise Only 05:03 Ambient vs Non-Ambient 05:43 How Ambient Mode Works 06:38 Support Email Use Case 08:32 Multiplayer Mode With Your Team 09:19 Shipping Career Site Filters 13:59 Website Security And Social Monitoring 15:34 Hooks, Security Checks, And CI/CD

@petergyang Long live!

Go to Settings -> Preferences -> Appearance -> Dock icon and Codex lives on.