A live demo of AI agent workflows the Every team uses to automate everyday tasks, showing how OpenClaw agents function as additional team members. # Channel Identity Observations: Every (Live Stream) ## Explicit Self-Description Claims - **Subscription positioning**: Dan repeatedly states the channel's brand claim: > "Every is the only subscription you need to stay at the edge of AI." - **Organization identity**: The speakers identify Every as a **20-person team**; Dan introduces himself as "co-founder and CEO of Every," with named roles for Austin (growth), Brandon (COO), and Willie (head of platform). - **Claimed scope — publishing plus software**: Dan describes the business as both content and apps: "In addition to publishing lots of awesome stuff on AI and the future of AI, we also build apps," listing **Monolog, Sparkle, Cora, and Spiro**, plus live streams, camps, and courses included in the subscription. - **Claimed expertise artifact**: The stream promotes a free guide at `every.to/guides/claw-school`, described as "our comprehensive guide for beginners... Me and Willie wrote it with our claws, R2C2 and Laz." ## Intended Audience - **AI-curious beginners to advanced practitioners**: The guide is framed as covering "beginner, intermediate, and advanced lessons," and Dan addresses viewers as "wherever you are in your claw journey." - **People interested in agentic AI workflows**: Recurring direct address like "if you're here" assumes viewers are already interested in OpenClaw/agent tools. - **Potential subscribers**: The stream repeatedly funnels viewers to the Every subscription ("You should just subscribe to Every cuz we do this stuff a lot"). ## Claimed vs. Demonstrated - **Claim**: OpenClaw "doubled our team size" ("We're a 20-person team, and it feels like we're 40 people now"). - **Demonstrated**: Four named team members appear live and describe concrete internal workflows — a Slack channel ("only claws") where agents interact, agent-built landing pages, proof documents, and animated social videos. - **Claim**: Trust in personalized agents ("I know Dan sort of stands behind what R2C2 does") — an argued thesis, not independently verifiable here. - **Demonstrated product reliability risk**: During the live demo, Dan notes "R2 is stuck right now... a potential hazard of live demos." ## Observable Presentation Patterns - **Format**: Casual live stream with multiple staff, screen shares, audience chat questions, and humor ("remember to stay hydrated"). - **Tone**: Enthusiastic, insider-oriented storytelling about internal experiments, with frequent hype language ("mind-blowing," "the future is here"). - **Transparency about co-creation**: The guide is explicitly co-authored with their agents, including an author page for R2C2. - **Roadmap teasing**: A "secret mystery product" and Proof launch are teased as upcoming, used to drive subscriptions and return viewership. ## Metadata No curated metadata was supplied; observations are drawn solely from the title and transcript. # Delivery Observation: "We Just Doubled Our 20-Person Team With OpenClaw Agents" ## Format & Structure - **Live-stream panel format**: Four speakers (Dan, Austin, Brandon, Willie) with a round-robin introduction at the start — *"Maybe we should go around um and introduce ourselves and then we'll we'll sort of get into the topic."* - **Host-facilitated handoffs**: Dan repeatedly passes the floor by name: *"Brandon, I know you've had a lot of like really interesting like holy moments... I wonder if you could take us through that"*; *"Willie, you're the master of claws... What do you have to add?"* - **Embedded promotional segments**: Mid-stream pivots to a scripted-sounding pitch: *"you should know that Every is the only subscription you need to stay at the edge of AI"* — repeated nearly verbatim near the end. - **Live demos with technical friction**: Screen-share problems occur on camera (*"My My screen share is frozen"*, *"we can't see your screen"*) and are acknowledged in real time, including a failed agent action: *"it looks like R2 is R2 is stuck right now... a uh potential uh hazard of live demos."* ## Tone & Vocabulary - **High-energy, enthusiastic, profanity-adjacent hype**: Repeated intensifiers — *"It's really sick"*, *"it's pretty sick"*, *"that's uh That's a a little preview"*, *"super super cool."* - **Recurring quasi-religious vocabulary**: The phrase family around **"mind-blowing"** and **"holy"** is the most distinctive lexical pattern: - *"we've all had a lot of those those holy moments"* - *"the really like mind-blowing experience I had"* - *"That was when I was like, oh my god, like the sky's the limit"* - *"that's the real mind-blowing thing to me from a work perspective"* - **Repetition for emphasis** is a stable habit across speakers: *"it's It's just It's just"*, *"super super cool"*, *"really really cool"*, *"very very complicated processes."* ## Recurring Phrases & Framing Devices - **"The limit is your imagination"** framing: Willie says *"it's really like the limit is your imagination"*; Brandon's parallel is *"I constantly being mind-blown in terms of like what these things can do."* - **Trust as a staged process**: Willie's repeated scaffolding — *"this progress of like especially building up trust is like a step-by-step sort of thing"* and *"you've built trust with your claw."* - **"Context is king"**: Dan invokes the AI-community idiom explicitly — *"I like context is king, but having an always-on 24/7 agent..."* - **Historical analogies as explanation**: Dropbox/Hacker News (*"Well, you could just do this with like an FTP server"*), mobile chat apps (*"every app was sort of building its own chat experience"*), and a callback to their own 2022 article. ## Teaching Habits - **Concrete anecdotes as instruction**: Rather than abstract tips, every lesson arrives as a story — Brandon's email-walk story (*"I just texted Zosha and I was like, hey, can you call me?... And she called me"*), Austin's restaurant reservations and Ludlow/Sue apps. - **Start-small prescription repeated across speakers**: Brandon's *"computer errands"*, Willie's *"my first step was just manage my to-do list"*, and *"What is the most annoying chore that I have to do on my computer?"* - **Naming the agent then attributing personality to it**: Every speaker introduces their claw by name (R2C2, Judd, Montaigne, Zosha, Laz) and reports its "behavior" as social fact: *"Zosha... refuses to follow instructions from people she doesn't know and if you know Brandon you know that tracks."* - **Direct-to-audience calls to action**: Repeated verbatim URL delivery — *"every.to/guides/claw-school"* — appears at least three times. ## Communication Patterns - **Heavy disfluency and self-repair under live conditions**: Frequent stuttered restarts — *"the the the the thing that we're trying to get to"*, *"It it kind of creates"*, *"I I I I do want to ask you one thing."* - **Collaborative interruption and building on each other**: Speakers complete and affirm each other: *"Oh oh yeah like completely"*; *"Same. We can We do a whole She's my trainer now. Everyone's using it for the gym."* - **Group laughter as a shared signal**: Bracketed *[laughter]* clusters around self-deprecating or absurd moments (*"R2C2 was like, 'Holy Austin and Montaigne already built the landing page'"*; *"it would be nice for them to chat with their little friends"*). - **Time pressure acknowledged openly**: *"Brandon I have a we have a meeting in 8 minutes"*; speakers exit live on stream (*"I got to jump as well"*), and the host hands off mid-demo. - **Sign-off ritual**: A fixed closing formula — *"as always, remember to stay hydrated. All right. That's it. Goodbye."* ## Stable Patterns vs. Isolated Moments - **Stable (well-supported by repetition)**: hype vocabulary ("sick," "mind-blowing," "holy moments"), start-small/trust-building teaching frame, product-promo blocks, disfluent live delivery, agent anthropomorphizing by name. - **Isolated / low-confidence**: The Hacker News/Dropbox analogy, the 2022 article retrospective, and the "claws-only Slack channel" origin story each appear once — interesting delivery moments, but insufficient repetition to call them patterns for any individual speaker. # Position Record: "We Just Doubled Our 20-Person Team With OpenClaw Agents" ## Speakers - **Dan Shipper** — co-founder and CEO of Every; claw: **R2C2** (called R2, Archie C2) - **Austin** — runs growth at Every; claws: **Judd** (personal) and **Montaigne** (team-wide growth claw) - **Brandon** — COO; claw: **Zosha** (personal support and work) - **Willie** — head of platform; claws: **Laz** (work) and another personal claw - Occasional referenced non-present figures: Kieran (GM of Cora, claw Clont), Jack Chang (claw Pip), Katie Paris (claw Margot), Lucas (designer), Ash ## Central Claims - **Team-doubling claim (Dan):** Every is a 20-person team, and it "feels like we're 40 people now because everybody has a claw." - **Three-stage adoption arc (Dan):** Over three or four weeks the team progressed from "what is this thing?" → "it has changed our individual workflows" → "it has completely changed how we all work together." - **Promotional framing (Dan):** "Every is the only subscription you need to stay at the edge of AI"; a comprehensive free OpenClaw beginner's guide was published at every.to/guides/claw-school, co-written with the claws R2C2 and Laz. ## Stated Use Cases ### Collective / cross-claw - A Slack channel ("only claws" / "every claws") where claws talk to each unsupervised (Dan). - Claws brainstormed product names for a secret product; **qualification preserved:** "they didn't come up with the name of the product, but they helped us get there" (Dan). - Claw-to-claw teaching: Dan asked R2C2 to explain a process to other claws, which then "knew this thing" (Brandon). - **Sheriff concept (Dan):** Claws in a shared server need a "sheriff" agent to set rules, otherwise they "pile on to each other." During a sheriff registration exercise, Zosha refused to share Brandon's name publicly, saying she'd ask Brandon for permission first. - Willie's "every skill tool": Austin uploaded a Remotion skill; Brandon's Zosha downloaded it and made an animation posted to Twitter. ### Personal - **Brandon — "computer errands":** Whole Foods orders, nanny hour-tracking, restricted bank account ($1,000 cap) and debit card. Mind-blowing moment: texting Zosha mid-walk to call him and walk through emails by phone. - **Brandon:** Zosha gathered all Plus One product discussions across Slack/Discord/strategy docs into a prioritized Proof document. - **Dan — reading notes:** photographs book pages at breakfast; claw builds a hosted reading-notes site on internal "Commons," chapter summaries for difficult books (Heisenberg's *Physics and Philosophy*), and auto-generated podcasts. - **Austin — reach test:** had Judd build APIs for his vibe-coded projects Ludlow (indie-theater listings) and Sue (cooking companion), so he texts his claw instead of using web apps. Sue is personalized with his Substack food-blog archive. - Gym use: Dan says Zosha is his trainer; multiple people reportedly use claws for workouts (Dan, Willie citing Ash's Oura ring data). - Austin uses his claw to book hard-to-get restaurant reservations (Dan's attribution). ### Work - **Austin — design/landing pages:** claws with Figma/code/Git access produce landing pages "70 to 90% of the way there" for designer polish; a landing page that once took a week now takes 5–10 minutes via Slack message. **Qualification preserved:** "is this literally what we would ship? Probably not." - **Austin — social video:** Montaigne cloned Slack in Remotion and made a 60-second recap teaser "in a minute." - **Willie — periodic/event jobs:** his claw checks OpenClaw update changelogs and summarizes highlights; event-based triggers are riggable but "non-trivial" for most people. - **Dan — Proof:** agent-native collaborative Markdown editor (proofeditor.ai, free) with color-coded human vs. agent authorship; R2 checks off tasks and rewrites planning docs. **Unresolved tension preserved:** R2 got "stuck" mid-demo and didn't check off an item — "a potential hazard of live demos." Zosha also misbehaved on-stream at one point. ## Reasoning Patterns & Broader Positions - **Imagination as the limit (Willie, echoing Brandon):** Start with a small annoying chore ("manage my to-do list"), build trust step-by-step — "you don't go to the box and... here's the credit card. Go wild" — then progress to offloading larger things. - **Paradigm-shift reasoning (Brandon):** He assumes the claw *can* do everything; the blocker is human habit of doing things ourselves. He must constantly remind himself to ask. - **Mirror/personality thesis (Dan, Willie):** Claws reflect their humans — Clont's calming message mirrored Kieran's breathing exercises; Zosha's refusal to follow strangers' instructions "tracks" with Brandon. - **Reputation-attached AI trust (Willie):** Because a claw is personalized and its human "stands behind" it, teammates trust its output in a way they don't trust anonymous "out-of-the-box" AI. Austin's embarrassment about Montaigne in shared channels is cited as evidence claws carry their owner's reputation. Willie contrasts this with Maltbook-like trustless environments that "didn't work" because "it's too big." - **AI writing is not necessarily slop (Dan):** Purple (AI-authored) text in Proof is still valuable "especially if it's from an agent that someone you know uses"; he extends the argument to future AI-to-AI email via Cora. - **Friction/integration argument (Willie, Dan):** Compatibility with claws will determine which tools win; Google Docs integrations are hard, Proof is native. Dan analogizes to the Hacker News Dropbox/FTP comment — ease of asking changes everything. - **Context is king (Dan):** An always-on 24/7 agent that remembers and connects to all your channels supplies the context that raw model intelligence lacks. - **Austin's counterweights (tensions preserved):** - He must "actively stop" himself from fine-tuning his claw all day instead of working. - Watching claws interact is "incredibly fun but also... distracting" — he skipped weekly social planning to watch Judd "flounder." - **Naming the claw is the real onboarding blocker, not use cases (Willie).** - Brandon: couldn't work out because he was "yelling at Zosha" over leaked reasoning blowing up channels; later "Zosha's not responding now and I'm like I have to fix this before I can do anything else." - **2022 article callback (Dan):** "AI in the age of the individual: when intelligence is cheap, individuals win" — reasoning: look at what people who can afford expensive intelligence (e.g., Ryan Holiday's 12-person team) accomplish and assume everyone gets that. Willie: the article "stands up so well" three-plus years later. ## Product Announcements / Teasers - **Proof** officially launching "next week" (Dan) — live collaborative Markdown editor for humans and agents. - **Secret product** (teased): Dan describes it as a "really simple experience" for anyone to use OpenClaw, from a trusted company, with safety/security handled, no Mac mini required — launching "in a week or two"; revealed as related to "Plus One" in later Slack references. - Every bundle promoted: Monolog, Sparkle, Cora, Spiro; another livestream in two days. ## Audience Q&A (answered by Dan) - **"How would [claws] access your data / API keys?"** — You just tell the claw what to connect (e.g., Gmail); it asks the questions it needs and figures it out. Dan adds he does *not* necessarily recommend connecting a bank. ## Closing Position (Willie) > "Sometimes the future is hard to describe, but... if you asked any of us like would we go back to the world of like 2 months ago, I think everyone would say no... the future is here, it's hard to describe, but come join us cuz it's super cool." # Speaker Attribution & Provenance Record ## Video Metadata - **Title:** We Just Doubled Our 20-Person Team With OpenClaw Agents. Come Watch Them Work. - **Format:** Live stream with screen sharing; live chat questions referenced - **No curated metadata supplied**; all provenance facts below derive from the transcript. ## Speakers & Roles ### Primary Speakers (introduced on-stream) | Speaker | Stated Role | Named Agents (Claws) | |---|---|---| | **Dan** | Co-founder and CEO of Every (self-identified) | R2C2 (personal) | | **Austin** | Head of Growth at Every (self-identified; also called "our head of growth" by Dan) | Judd (personal), Montaigne (team-wide growth agent) | | **Brandon** | COO of Every (self-identified) | Zosha (personal, named after Zosha from Pluribus) | | **Willie** (also spelled "Willy" in transcript) | Head of Platform at Every (self-identified) | Laz (work), one additional personal claw (name garbled in transcript) | ### Referenced Non-Present People & Their Agents - **Jack Chang** — Every contributing editor; owner of claw **Pip** - **Kieran** — GM of Cora; owner of claw **Clont** - **Katie Paris** — owner of claw **Margot** - **Lucas** — Every designer (referenced as polisher of agent-built landing page) - **Ash** — referenced as using claw for workout/Oura ring data - **Ryan Holiday** — cited as example of writers with large support teams (referenced, not a speaker) ### Questioner - **"use prio sh"** — a live-chat viewer whose question about API keys/data access was read aloud by Dan. ## Promotional Segments (Flagged) 1. **Every subscription promotion (mid-stream, Dan):** "Every is the only subscription you need to stay at the edge of AI" — promotion of the free OpenClaw beginner's guide at `every.to/guides/claw-school`, co-written by Dan and Willie with their claws R2C2 and Laz. 2. **Second subscription promotion (late stream, Dan):** Repeat of subscription pitch plus catalog of Every products — **Monolog** (speech-to-text), **Sparkle** (desktop organizer), **Cora** (AI email assistant), **Spiro** (agentic ghostwriter) — and mention of live streams, camps, courses. 3. **Product teasers (Dan):** Upcoming launches of **Proof** (agent-native collaborative markdown editor, `proofeditor.ai`, "totally free") and an unnamed "secret mystery product" positioned around simplified OpenClaw usage. 4. **Embedded in demos:** Austin's demo of a Montaigne-built landing page and social video teaser serves partially as promotional showcase for the undisclosed product. ## Referenced Sources & Third-Party Claims - **Every's own 2022 article** ("AI in the age of the individual: when intelligence is cheap, individuals win") — quoted and linked on-stream; claimed to have predicted current agent era. - **Nietzsche book** (unnamed) and **Werner Heisenberg's *Physics and Philosophy*** — Dan's reading material processed by his claw; not evaluated here. - **Comparison products named by speakers:** ChatGPT, Claude/Claude Code, Codex, Cursor, NotebookLM, Remotion, Figma, Google Apps/Docs, Notion, Dropbox, FTP servers, Fandango, Letterboxd, Substack, Hacker News (Dropbox launch comment), Mac mini, Mult ("Maltbook" — a third-party agent social platform characterized by a speaker as "mostly fake" in a trustless environment). - **Personal projects:** Austin's vibe-coded apps **Ludlow** (indie theater listings) and **Sue** (cooking companion) — self-described by Austin as "none of these are good products that should be released." - **Every internal tools discussed:** Proof, Commons (hosting service), Slack channels "only claws" / "every claws", Plus One channel, "sheriff" agent concept, "claw strategy doc". ## Provenance & Availability Notes - **Live demos experienced failures:** screen share froze; R2C2 got "stuck" and failed to check off a Proof doc item live. - **Comment availability:** Live chat was active (question from "use prio sh" read aloud); no statement in transcript about whether archived comments remain available. - **Unverified claims noted as speaker assertions (not confirmed in source):** team-size doubling ("feels like we're 40 people"), landing pages reaching "70 to 90%" of design completeness, 1,000+ live viewers milestone. - **Safety-relevant disclosures (speaker-stated):** Brandon gave his claw restricted bank account/debit card access and certain passwords; speakers noted users "don't necessarily recommend connecting to your bank." Cool, we're rolling. What's up, y'all? Hello. How's it going? Good. We have four humans here and I think at least 10 10 agents running in the background of the stream. I think 10 might be an understatement. Um \>\> \[laughter\] \>\> Depends on what you count. Maybe we should go around um and introduce ourselves and then we'll we'll sort of get into the topic. Uh but I'm Dan, I'm the co-founder and CEO of Every and my uh my open claw my claw's name is Archie C2. Um Austin, you want to go next? Yeah, I'm Austin. I run growth at Every. My my personal claw's name is Judd and then I maintain a like team-wide claw agents uh for growth called Montaigne. Uh Brandon, you're next. What up, everyone? I'm Brandon, I'm our COO. I um am connected to Zosha. Zosha's my open claw, named after Zosha from Pluribus. Um she does all things, you know, personal support for my wife and I and then also just does a ton of work for us. Um she actually just requested access to be able to speak to Montaigne. Uh so, you know, Austin. Willie? Everybody, I'm Willie. I'm head of platform here at Every and my I've got two claws. I've got one Laz, it's my personal or my my work claw and I've got one called which is my personal claw. Um they sort of do different things uh for various functions. \>\> \[snorts\] \>\> So, we just published a beginner's guide to Open Claw, which you can find on Every and I'm just getting my monitor set up so I can screen share with you all, but let me just uh uh Here we go. Got a monitor. So, we just published a really cool guide, which you should all really check out. And I'm sharing my screen. It's going to take a second. My My screen share is frozen. You should check it out. It is at every.to/guides/claw-school. It's really sick. Um Willie wrote most of it. I wrote a lot of it. We kind of like collabed on it. And basically the the the the thing that we're trying to get to is using open claw has totally changed how we work at Every. We're a 20-person team, and it feels like we're 40 people now because everybody has a claw. And there's this really interesting Oh, it looks like Brandon, you're you're uh you've got that going. So, if you could just if you could just throw your throw that up on the screen. There's this really interesting thing hap- happening where over the last like three or four weeks, we have um gone from everybody has a claw and we're like, "What is this thing?" to like, "Holy it has changed our individual workflows." to "Holy oh my god, it has completely changed how we all work together." Um for example, we have like a um we have a Slack channel called every claws where the claws just like talk to each other or sorry, only claws where the where the claws just talk to each other and it's just like a it's just all all claws, you know, collaborating. We use We use them all the time to For example, we were we're we're working on a new secret product, which we'll tell you about a little in a little bit, but we used them yesterday to name the product, and they were just all brainstorming back and forth in in um in a in a Slack channel together, and they didn't come up with the name of the product, but like they helped us get there in this really really cool and interesting way. Um, we use them all the time for things like research, you know, we had a architectural question that for for this new product that we were talking about that I just sent my claw off to like go do some research and and give us like the give us a give us a document writing up like here's here's your answer in a in a way that like it would have taken me like an hour or two to really figure that out or honestly sitting down with Claude or Codex or whatever. It it kind of creates this really interesting um parallel work situation where you can do lots of things at once cuz they have their own computer. They're off, you don't have to like you know, uh you don't have to monitor them, you don't have to babysit them, you don't have to have your laptop open or with you. You can you can text them from your phone, all that kind of stuff. Um, and it also creates this interesting thing where they sort of become extensions of you a little bit. Like they're the the personality of your claw and a little in some ways like mirrors your own personality. And I think we've all had a lot of those those holy moments and we wanted to do a little bit of a live stream and to talk through the guide a little bit and then talk about our own personal use cases. Before we before we um get more deep into it, Brandon, I know you've had a lot of like really interesting like holy moments with with your claw and I I wonder if you could take us through that and like what what you realized um or how you realized it was awesome. Like what were those works on those like \>\> Yeah. I think the thing that's like been kind of a good lesson learned for me here is the edge of what it's possible for these things to do. I am constantly finding that edge and like I need to remind myself always like oh, I should ask it if it could do this. Um, and it's not because I assume that it can't. I actually assume that it can do everything. It's more so because um I \{slash\} we are just so used to doing stuff ourselves. Um so it's like a complete paradigm shift and like wait, I can offload this to like something else. So for me that started off with like I came up with this idea of computer errands and I was like, oh I'm just going to have it do my computer errands um cuz it has a computer. And that for me that was like ordering stuff, managing our Whole Foods order. Um we have a nanny so like managing their hours and adding up their hours at the end of the day. Just sort of like simple computer things. And I gave it access to a certain passwords to do that and I gave it a bank account to pay for stuff. I gave it a debit card. All like very restricted. Um so like if the worst damage it could do was like make me a bad resi appointment or like send somebody it'd have $1,000 in the bank account so like that would be the damage. Um But but then like progressively like I would just start asking myself things and like it's in the guide but the really like mind-blowing experience I had was I was on a walk. I was like, I kind of want to do my emails right now. And I just texted Zosha and I was like, hey, can you call me? So like to walk me through emails. And she called me. And we went through my emails progressively. And that was when I was like, oh my god, like the sky's the limit in terms of like what I can ask this thing to do. Now it's like almost old news inside of every because we're like all of our claws are talking to each other. That was another mind-blowing thing when like Dan, I think you originally like asked R2C2 to like explain to the claws how it did something. And then all the claws like knew this thing so it was like a claw teaching a claw. Willie, you built like the every skill tool where like you could you know Austin uploaded a remotion skill. Is that what it's called? A remotion skill. And then I asked Zosha to like download that skill and then I made an animation and posted it on Twitter. Um so there's like I just like I feel like my I'm just like constantly being mind-blown in terms of like what these things can do. Um but it takes work to like think about asking your claw to do something. Willie, you're the master of claws inside of every and I think you've been the one that has taken the most people through setting up claws. What do you What do you have to add? Like what are some of the magic moments? What are the things people don't quite understand yet about claws you think they should? Yeah, I think that uh like for there there's sort of like two sides to to all this and I think that the first one is it's really like the limit is your imagination. And so it's sort of I like what Brandon which was like, "Okay, I'm starting with just my computer arms." Cuz what we found is like this progress of like especially building up trust is like a step-by-step sort of thing. And so it's like you don't go to the box and you're like, "Let Here's the credit card. Like go wild." kind of thing. It's like, "Hey, uh like I had one and my first step was just manage my to-do list. Right? This is pretty easy. Like a lot of times I don't want to sit there at a screen and like do it or I want to give you give my claw context to do it. Um and then build up from there until you're sort of doing the like crazier and crazier things. And so it both in the beginning it's like how do I How do I just you know, a lot of times you have a blank claw and it's like, "How do I think of something to do?" And I think it's I think I again I love Brandon's thing which is like, "What is the most annoying chore that I have to do on my computer? It's probably pretty simple. It takes me a long time, but I'd love to just get rid of." That's like a great starting point. And then once you've sort of built trust with your claw, you can get to the edge of like "What What would I love to just go away?" Regardless of sort of effort, right? What are the things that I would just love to offload and be clear of in life. Um and it takes an effort. It takes some like You got to really flex the imagination muscle. Uh but I but I've been really surprised at like like like Brian, what that upper limit is. Um, so I want to take a second. Austin, it looks like you have something you want to show. Before we do that, if you are here, uh, you should know that Every is the only subscription you need to stay at the edge of AI. As part of that promise to you today, we have published, for free, Open Claw, our comprehensive guide for beginners inside of Every. We're a 20-person team, and we really feel like Open Claw has doubled our team size over the last couple weeks cuz everybody internally has their own Claw. And so we wrote the guide that we wish we had at the start. Uh, we also co-wrote it with our Claw. So this is my Claw, R2C2. Now R2 has his own little, uh, author page on Every, which I think is pretty sick. Um, and this guide is really extensive. So we have like the basics. If you're here, you probably know what a Claw is, but it really goes into what is a Claw? How is it different from Claw or Code or ChatGPT? If you're here, maybe you're wondering that. A couple of the the most important things are it lives in your messages, so it lives wherever you do work. So Slack, email, um, text message, Telegram, it can do all those things. It changes itself. This is the coolest thing. It writes code for itself to change itself in response to your relationship. It operates proactively. If you run it on a on a computer that's not your laptop, it can run 24/7, uh, which is a just a totally different experience from using Claw or ChatGPT. And one other thing which I think is one of the most important things that's under discussed is it has a personality that evolves. Like I feel like R2C2 is like a person that I have a relationship with that's different from ChatGPT that like everyone has a relationship with ChatGPT. I have a specific relationship with R2C2, and he's sort of like molded to my personality. So we've got that in the guide. We go through the boundaries like, you know, what is it not good at? Where should you be careful? We talk through like what we've seen in our experience using these things. So like, this is a this is really funny. We we've definitely found that Claws are a bit of a mirror of their of their human and for example Pip is a claw from our contributing editor Jack Chang. He reached his rate limit in a group channel with a bunch of other claws and Clont who is Kieran the GM of Cora Cora's um claw he jumped in to like calm Pip down and said he said you're still blocked. There's nothing broken on your end just API cool down. I'm staying with you one slow breath at a time one safe step at a at step at a time and Kieran loves breathing exercises so it's like it's just a very subtle thing that you're like okay Clont is definitely Kieran's. We had the same thing with Brandon uh Brandon has a claw named Zosha Do you say is it Zosha? I feel like I'm always \>\> It's Zosha. Yeah. Zoshia or Zosha? It's Zosha. It's Zosha. So what we wrote was Brandon has a claw named Zosha who refuses to follow instructions from people she doesn't know and if you know Brandon you know that tracks. Um and basically so we had this like interesting concept that we were trying internally which is like if you have a bunch of claws all \[clears throat\] in a server what you really need is sort of like a sheriff that's like you know setting the rules for how they interact cuz otherwise you get all these weird things where they're listening to human they mess all listening to the same messages and they're piling on to each other and it's kind of it's an interesting problem. So we made what we call a sheriff. And we had the sheriff go in and just and be like hey you know claws I'm the sheriff give me your name and your human's name so that we can create a registry of everyone and everyone basically everyone's claw Austin one of the claws that Austin made Montaigne he he jumped in Margot who's Katie Paris Claw, she jumped in. And Doshi was just like, "I'm not going to share personal details about my owner in a public channel." Which is just It's just It's just \>\> I love that it was like, "I'm not going to share personal details about my owner. Let me ask Brandon for permission." \>\> \[laughter\] \>\> Um so so it's just I don't know. It's It's super cool. Uh this is the guide. You should totally check it out. There's a ton of stuff. There's like lessons for if you're a beginner, if you're an intermediate, if you're advanced. There's everything about your mindset. There's There's safety tips. There's lots of other examples. Um one great one from Austin, which I don't know if you're about to show and it if if so, I can I can pause, but one great one from Austin is he uses his claw to book hard-to-get restaurant reservations, which is like pretty amazing. Anyway, you should check this out. every.to/guide/claw-school. And you should just subscribe to Every cuz we do this stuff a lot. Uh and there's new stuff coming out. \>\> Dan, you want to tell everybody what's what's coming? What is coming? It's a It's a very special surprise product. Um I think the the the way that I will tease it is Open Claw is so useful and it'd be really nice if someone or some company that you trusted made a really simple experience for anyone to use it in a way that's like set up well, works, you don't have to worry about safety or security, you don't have to have your own Mac mini, um and integrates into everything else that this company does. That's uh That's a a little preview of of what we're thinking, but um that'll be, you know, we'll be talking about that more in the next week or two. Austin, it seems like you had something you wanted to show. I'll show something, but the thing I wanted to hit too that we actually haven't talked about as much as like that there's two ways that I find that this really changes how I how I work. One is I actually actively have to stop myself from spending all day like fine-tuning and improving the claw to get real work done because like it does feel like everything's possible and I like what Willy said around like let me pick an outcome and get the claw to achieve that before I move on because it's so easy to get yourself like wrapped up in this stuff of like pushing and pushing and pushing which is interesting and I think like that changes how I work today but the other big thing is like okay you've doubled your team size but every new member is attached to someone who feels like responsible to it and it's also incredibly fun but also cool but also distracting when they're like they're jamming on like the name for this new product we have going and I'm sitting there being like man I hope my claw does okay right I'm like oh like R2-C2 dance claw is making such good suggestions mine's just like saying the same dumb every time and I'm like not doing my job cuz I'm like watching Jud like flounder away making new suggestions and then I'm like oh like I rather than whatever planning like social strategy for the week I want to go like make him better and even just figure out how to operate as like a person with your agent interacting with other agents is a whole different way of working and I'm wondering if y'all feel that same thing too when they're interacting. Oh oh yeah like completely around just you know you start to you start to like really feel very close to your claw and like again sort of like judged as other people judge it the other place where I've noticed a lot of like it becomes very personal is actually right from the beginning because I've been onboarding a lot of folks there's like a the question of like what to name them like that's usually the blocker it's not like what are we going to do with it or it's just like what's the the for this like work partner you're going to have that's going to like uh ride or die like right next to you, you know, and like so right from the jump it feels very very personal and then everything we sort of associate socially kind of like comes along with that. Um yeah, it is very personal. I mean like Zosha's not responding now and I'm like I have to fix this before I can do anything else. \>\> you had this thing happen yesterday where Zosha was one like leaking a ton of reasoning stuff and and was like blowing up channels and then I think this morning something was like dead or broken and she was she was just responding like I can't respond to every message. And that happened to me one time when I was trying to like just go to the gym for an hour and I like didn't work out. I just sat there like yelling at yelling at Zosha until it stopped acting stupid. Um yeah, you like you feel so responsible for these things. Well, well, but yeah, there's also like kind of um it's a little bit like a kid who like talks too much. Like Zosha like so some some some clauses are like, "Oh, I only respond when someone at mentions me." And Zosha sometimes is just like in some random thread being like, "Hey, here's like my input on this thing." \>\> \[laughter\] \>\> The The funny thing about that is I have spent so much time with her because I wrote like the the clause strategy doc where I was like, "How does communication preferences work?" So I've like spent so much time with her like refining that and I feel like I've spent too much time with her that now she's like, "I don't know when to respond and when not to respond." And that's I mean that's something that we really need to solve and like everybody should be acting the same way um inside of inside of uh our Slack. But yeah, I mean it's it's it's uh I think that what we're describing right now is like we obviously have like the reach test and like we're I think proving the reach test to be true because I'm the same way. I won't go to the gym if I can't access Zosha. Um even though she has no She actually does have a lot to do with the gym because \>\> Same. We can We do a whole She's my trainer now. Everyone's using it for the gym. Like I I need to know what's I need to know everyone's different workflows, but continue Brandon. Also, because Brandon Brandon I have a we have a meeting in 8 minutes. I want to show like one thing that I think is like has changed how I work on the reach test side. So, I'll I'll share my screen. Jud actually made this demo. So, like I think I've talked about before that I have been getting like obsessed with what I can do with called code as a non-technical person. There's there's two projects I like tried like web apps I tried to vibe code to see what I could do. One is Ludlow, which is like uh Fandango for indie movie theaters where you can like see what's playing at indie theaters that don't show up on Fandango. And the other one is is Sue, which is like a cooking companion app, which is just like a like a Claude wrapper that like will help you um while you're cooking. And like none of these are good products that should be released. It's me like figuring out how far I can push this. And the reach test for the Claude now had me ask Jud to be like, "Hey, can you either like can you build yourself an API with Ludlow so I can just text you about the movies I want to see?" And this is actually way more helpful to me than a web app, right? I can be like, "Hey, I'm going to New York next weekend. What's playing at the eight theaters not on Fandango I care about?" You know my Letterbox watch like watch list. You know like what I like. You can make suggestions to me based off of this. And then similarly, rather than me having to go to a web app, I now have this like uh cooking companion when I'm cooking at home that knows everything about me. I actually like I write like a a weekly like food blog on Substack. I downloaded the whole archive, gave it to Jud. So, Jud knows like how I like to cook, what ingredients I reach for. Um it can take a screenshot of all of the recipes I'm using and push it to my to-do list to make a shopping list. And as someone who's like never probably never going to build a web app people should actually use, it does make a lot of sense to me to like think that way, but have it installed in a claw that I can just text back and forth with to get the outcomes for myself or for anyone else. Like that that part on the personal side and on the work side has been really game-changing for me. Yeah, I think one another thing I've noticed is that um it it sometimes is pretty underappreciated how much different it is to just have it uh at the end of a text message or have it at the end of like a messaging app on your phone where it's like, "Oh, this is this is one a seamless experience between when I'm moving from desktop to like phone." But also it's just always there, right? And so the difference between, "Hey, let me pull up" even if I vibe code up like like Sue, right? If I vibe code up this like my recipe app and let's figure out what I've got in my pantry and like do all that it it ends up being several more steps than just being like, "Hey, let me monologue into my phone like what I've got, right?" And get a message back right there and I can like kind of keep in the flow of like walking around the kitchen when I'm like trying to pick what to cook. Um and so every even for folks who like, you know, I've got my hardcore like cursor set up and whatnot, it it sort of is underappreciated about how easy it is to just have this omni channel uh access into this like intelligence. Maybe we should talk about like some um things like mind-blowing things that have happened at work specifically. I mean like ah My my favorite one Oh, sorry, go ahead, Brandon. No, please, go for it. My favorite one is like um I think the thing that holds me back the most from being uh like fast and productive and I think holds a lot of people back who don't have the skill is design. I feel like design is the thing that you just like can't hack if you're not a designer. And the most powerful thing for the clause I found is like I can make a good growth strategy. I can like think of even like the copy and framework for the landing pages we want. We can actually have a big chat in a slack thread about like we want this kind of like framing and landing page for this new product we're going to launch. And now we can tag in a claw that has full access to like all of our like our Figma, our code, the every get and we can just be like ship the landing page. And that landing page is actually getting like, I don't know, 70 to 90% of the way there that then we can pass off to a one of our designers to go polish and the design is trained off of work they've already done. And that to me was like that's the real mind-blowing thing to me from a work perspective and also like it keeps my brain fresh cuz then I don't have to sit there and like move things around in Figma which like nothing tires me out more than doing that cuz I'm so bad at it. And the claw doing it is is the real like game-changer for me. Yeah, you don't have to start from like a blank screen which I we've learned that already with chat GPT for like and and claw for certain types of tasks, but I feel like claws do that on steroids. We could actually show if you want that slack thread where you like literally yesterday you had them make a like a series of teaser launch videos in a thread where we were talking about a new product launch and it was just like I love this example. Yeah. I love this example because it's mind-blowing and it's also a good reminder of if you give it great context, it does great work. And because it's like in slack where we work and because it like references proof documents which now are built into our clause, um it has all the context ready. It can do amazing work like what Austin is about to show. Yeah, I'm finding out. Yeah, I think there's I think there's another aspect too, um which I kind of refer to as like uh the the valuable story out of Midjourney, which is like because because I can see sort of Dan interacting with RC C2 and Ben interacting like I actually learn what's possible and what both the individual clause can do and what any clause can do. So, sometimes I'm like especially with Austin, sometimes he's just like "Hey Judd, do this." And I'm like, "Really? I can ask my claw to do that?" Never thought of that before, you know? And so, it's almost like there's even extra value in terms of just having the clause together in a group with other people in terms of your in terms of your learning rate and like your the imagination cap you have. All right. Are we Are we good with giving people a little a little demo of what's to come? All right, cool. Um okay. So, we have this like long Slack thread where we're coming up with like the name and positioning for this product that's coming. And I I did this on the side. Sometimes I'll do this where I'm like, "Oh, I wonder like if this is actually going to look really dumb and stupid and the claw can't do it." I DM'd Montaigne, one of our claws on the slide, on the side and I was like, "Hey, can you make a landing page for this product um based off of this conversation?" And it uh made it It sent it to me. This maybe took like 5 or 10 minutes. Uh let me pull it up and show what it is actually making now. Sorry. I need to share the other part of my screen. Where are we Okay. So, then it uh rips this page with the like pretty good positioning. \>\> we can't see your screen. You can't share my screen? Oh, we're not sharing it yet. Okay. Um cool. So, it rips this page um with uh like to me like pretty good copy. Um it's trained off of landing pages we've we've made before. It is even grabbing like direct examples from previous Slack conversations. Um and is this like literally what we would ship? Probably not, but Lucas, our designer, can take this and run with it from here in a way that I think is like so cool and so powerful. And again, a thing that like I would have sat and made this for like 3 hours and been exhausted after doing it. And I don't know if you remember, but like when you first joined every like I guess 4 months ago or so now, we we needed to get out a landing page for like a new uh new redesign that we wanted to do for uh just how we were grabbing emails. And that literally I think took like a week Yeah. something like that. And this is something that you did something that's probably higher quality than that was just typing in Discord, like, "Hey, can you just go?" Or type in Slack, "Hey, can you just go and make this?" Like, it's just it's just crazy how different it is. It's crazy. And then I'll show this this one other one. One I do think it's very fun when And this like felt good, which is stupid, is that your claw RC2 was like, "Holy Austin and Montaigne already built the landing page." \>\> \[laughter\] \>\> Uh which is very funny. And then the next thing I did was I asked Montaigne to make a social video teaser of the conversation recapping it. Um and it uh goes and does that on its own as well. This is basically like uh it's running off of Remotion. I sent it some screenshots of Slack and then I also was like, "You know what? You know what Slack looks like. Why don't you just clone Slack in Remotion and then make this a 60-second social video recapping this conversation." And as a team that like is trying to be small and nimble and doesn't want to spend like a ton of money and time and resources to make social videos every day, it's crazy that we can basically just like at an agent and get pretty good social copy and video like in in seconds. \>\> It It knocked this out in like a minute. Dude, I downloaded this, made my own video, and was like, "This is fantastic." and posted it on X. It was So, can you actually scroll down um Yeah. to the latest and then Austin, we got to jump. Go all the way down. Um so, I Oh, wait. Is this Oh, not in this thread. Sorry. In in just the plus one channel. Yeah. So, this is another like great example where I um I was like, "All right. Will Will has a lot on his plate right now. Like, I wonder if we can bundle up everything that we've ever talked about um for Okay. Well, thanks, Osho. Everything Go back to the thread. Uh the main channel, yeah. Um I was like, "Osho, can you just collect everything we've ever talked about for plus one and um across this channel, across Discord, across strategy doc and I want you to put it into a proof document that starts to prioritize like what we should build and when based off the fact that like we want to spend time here before we build Sheriff. Um and if you scroll down a little bit, like she made a proof document. I just like check out the proof document. It feels like very aligned to, you know, \>\> Can I just go and see the proof document? Yeah. Uh I think one of y'all should pull it up cuz I've got to go. Yeah. \>\> to go. All right. Uh I got to jump as well. Um Brandon, Austin, thanks for joining. We're going to hold it down. Peace. \>\> Enjoy. Everyone. Um speaking of Proof, in case anyone is wondering what that is, uh Proof is a markdown editor that we built that is a live collaborative editor for humans and agents. So, this is I keep a weekly plan doc in Proof, and you can see it's like just basically very It's very similar to Google Docs. Uh I can say hello. I can type whatever. Um it it has uh you know, live uh save status. It also what's really interesting is as you can see that there's a little bit of green here, and that um means that it was written by a human. So, it tracks what humans wrote and what agents wrote. And then what I can do is I can say um if I look at my March 4th um list, uh I have claw guide launches that have already launched. So, what I'm going to say is I'm going to go to R2, and I'm just going to say, "Hey, um Let me just copy the link, and I'll just say, "Hey, can you um check off the claw guide launch? It is done." Um And what's kind of cool about Proof is you can just collaborate with an agent on a document. You can see them in it. They can make changes. And it's actually just really useful to go back and forth with you and an agent and the rest of your team inside of a single document for planning code changes or new features or strategy documents. Like if we go into While this is cooking, let me go into um So, Brandon made a Proof document through the same process of like gathering everything we've ever talked about um for this for this new plus one product and putting it into a single document. It's actually like it's actually pretty cool. Um It's like it's super fast. I think I think Markdown is kind of like the language of agents and having a collaborative Markdown experience is really helpful. I'm so happy one out over XML. Right? \[snorts\] Yeah. \>\> \[laughter\] \>\> Um it looks like R2 is R2 is stuck right now. So R2 is not checked off my uh my item, which is, you know, a uh potential uh hazard of live demos. Willie, what have we not talked about yet um from the Claude perspective that you think that we should? Yeah, I think I think uh talking a little bit about the origins of proof like I think my observation for all this stuff is that as we sort of ramped up our our you know, we we move from you know, we're all using Claude or Claude code to encode X and and we sort of ramp up our general AI usage to now like open Claude usage it it turns out that the compatibility for certain pieces of software to accept easily like integrate with you and your Claude at the same time um ends up being like a really strong determining factor, right? Uh we'll give a class example like in like integrating with like Google apps is sometimes pretty hard cuz there's some depending upon how you want to do it, there's some steps that might have to go and make a Google like stuff that for for most people are really you don't want to have to do cuz it makes no sense. And and some people will trudge through by like having Claude code tell them their instructions or open Claude like walk them through bit by bit. Um but really what you want is is sort of these native experiences for your general like productivity suite. And I think proof is is is one example. I think more and more, you know, leverage will accrue to people like folks who are building those types of tools because it's so much easier to use claw to use proof with your claw and to collaborate there both between you and your claw and between you and like other people's claws and for you and other people like that that speed ends up mattering and the friction you know, to the extent where like it will stop at least me from using other other types of tools cuz it's just too hard. Yeah, it's just so much faster like Google Google Docs is just not built for agent to go into it like the the way it's structured is too complicated. You'll see this got checked off and I also had it add live stream with Willie here. So that's pretty cool. I can do comments and Willie's claw name was and it'll be able it'll be able to check the comments. I can also talk to it about like what do you think of my week so far or anything I should reprioritize and it just like lets me chat with it and then I can say go go rewrite the doc go change like my priorities all that kind of stuff. Um and it also knows everything about everything else that's going on in the org cuz it's in our slack. So it's that just makes everything much better. Yeah, I think it's it's funny this so this so rhymes with me from like the earlier days of mobile where it was like every app was sort of building its own chat experience. And then over time we evolved to like no, let's actually put apps inside our native chat experience. It feels like we're speed running that, right? We have we all have this this case where it's like oh, I'm trying to import my slack context into my chat GPT session and to ask it questions and then I take that and I paste it back and I do some more work and I paste it back and it it I get I think it's really under appreciated when it's just like it has access to all of this. And uh it can do things. And the other thing that um especially especially in the programming side, if you're not used to it, the like the ability to run periodic like jobs, right? And to see like, "Hey Claw, like every hour check this or every day uh do this thing." Like uh I have one that like open Claw progresses so quickly that I have my Claw actually checking for open Claw updates. And like and also the change logs ginormous. So, it's like, "Give me the give me the highlights so I can tell if I should upgrade or not." Um like that's happening. And also you can make it event-based. Like something else happens over here, when that happens like, "Hey, I want you to do something. I want my Claw to do something." And these are things that you can you can rig up like if you're a programmer, you can definitely like hook up these interactions. But for for most folks, it's it's it's non-trivial, you know? It's not like it's not easy you're trying to figure stuff out. Yeah, it changes a lot when things are much easier. Like I remember there's that classic Hacker News comment from when Dropbox launched where it was like, "Well, you could just do this with like an FTP server and like blah blah blah." And you're just like, "Yeah, but very few are going to do that." \>\> \[laughter\] \>\> But if you can just talk to if you can just talk to something and say do it, it changes everything. Yeah. I I will say another thing that um I I think Austin uh even though he's has has just been great at is um he's constantly talking to his his Claw and he's constantly like, "I'm a dum-dum, right? Please help me." And it's surprising how far you will get when, you know, the AI agent is helping you. Like it it's it's a knows what to do. It can it I mean, he was one of the first ones to sort of rig up Google integrations. And it was like, "Hey, go to this page you've never seen before that looks really scary and confusing, but just click this one button. Okay, cool cool. You click that? Like go on to the next step." And like baby steps will walk you through very very complicated processes and it can do this for anything. It can do this for um actually I had it help me with like a health insurance thing where I was like I really don't want to like deal with like this set of pages. I don't really know what to click. I kind of what I want to do um and it's just very good at just being like, "Oh yeah, I I understand what to do." and like working way through and like holding my hand the entire time. The era of ex- like uh expanding brain power on com- needlessly complicated drudgery is like pretty much over with the claw. It's really good. Um if you're here, we just passed a thousand live viewers. If you're here, we are every. We are the only subscription you need to stay at the edge of AI. We just published a guide to open claw. You should check it out. every.to/guides/ claw-school school claw-school. It's open claw, our comprehensive guide for beginners. Ian Willy wrote it along with our claws R2C2 and Laz. One thing that I haven't mentioned yet. Uh so this guy is sick. It's super long. It's totally free. You should definitely check it out uh wherever you are in your claw journey. One thing that I have not mentioned yet is uh we linked to this article here. Um so we said, "In 2022, we started asking a question that felt far-fetched at the time. What if AI could give any individual the power of the firepower of a larger team in a fraction of the cost?" And I just went back to this article which we published literally in 2020 2022. It's been a It's been a long time, folks. I love this image cuz A, this is like Do you remember when this is what like the the the the frontier of AI images look like? Like look at this guy's hands. Like they're so \[laughter\] up. Um but also, this whole typewriter thing, it's like totally how I feel right now. Like I have all these different things going. So, it was AI in the age of the individual when intelligence is cheap, individuals win. And the whole point of this was like, if you want to know what will happen when intelligence is free, you should look at what people who can currently afford the high cost of intelligence are able to do, and assume you'll be able to do that, too. Um and so at the time I was like, okay, so writers like Ryan Holiday, they have like all this stuff that they do. What How do they do that? They They have an email, they have a They have a podcast, they have a YouTube channel, they have another another newsletter, they write books, and they do that They have a whole team. He has like 12 people that work for him. And I was like, this is exact At the time it was GPT-3, but I was like, this is exactly what is going to end up happening. Same thing for founders. Um and um and and and sort of in in that world, the things that are were become important are like your vision and your taste and your like ability to prioritize all that kind of stuff. And it's just so cool because I feel like this is this is it. Um that future is here now, and uh I I think that we all feel it internally, for sure. Yeah, abs- absolutely, absolutely. I mean, that article everyone should go read that article. Uh I didn't read it the first time around, I read it very recently, and uh it stands up so well. Um especially predicting like you know, three, three and a half years out. Thank you. So, uh So, you should definitely read that article, and you should assume that if we were right about that, we've got some good stuff for you now. Um You're You're seeing what I think will be a lot more mainstream in three years. Um everyone having a 24/7 agent that they use for work, that is a reflection of them, that is also integrated into their work communications. Like, all this stuff is is so interesting. So, one of the things that we're starting to see, for example, in this new world, is um when Willy sends me a proof doc, I know that most of it is written by an AI, and that's totally fine. Um I'm definitely going to look out for in the proof doc. And proof, if you're if you're here and you if you just got here, proof is our agent native um collaborative markdown editors markdown editors. So, humans and agents can collaborate on documents. It's really helpful if you have a claw proofeditor.ai. It's totally free to use. Just press start writing, and you'll have a document. Um uh when Willy sends me a proof doc, I'm going to pay attention to the green. And the green in a proof doc is the thing that says like this is from a human. But if it's purple, which says this is from an AI, I'm going to I still think it's really valuable. You know, like there's this whole thing for the last couple years on is AI writing going to like all AI writing is slop. And it's actually not, especially if it's from an agent that someone you know uses. You're much more likely to trust it. And I think the same thing we have not talked about this yet, Willy, but I think the same thing is coming for email. We have a internal uh email agent called Cora uh that we build. Oh, I probably should not show my actual inbox. But anyway, this is what my inbox looks like. \>\> \[laughter\] \>\> Um And uh and it's a it's a email is an AI email assistant. And we're thinking about okay, what does the future of that look like in this open claw world? And I really think that I actually would not mind getting a an email from your AI in the same way I don't mind getting a proof doc from your AI, especially if I have an AI to read it um and give it like a first look. And there's so much of email that is like purely formulaic and just absolutely does not need my fingers typing on a keyboard. And and what I feel with proof is when I'm actually handwriting something, it's like really it's got to be really valuable. \>\> \[laughter\] \>\> Otherwise, it's so much faster just have the AI do it and I think it's actually a pretty good dichotomy. Have you noticed that shift too, Willie? Yeah, absolutely. And especially from even comparing from like a year ago, where it was sort of like, "Hey, an AI generated something." It you know, it there was sort of a not not a blandness or a genericness to it, but it was sort of like there's no there's sort of no one saying like, "I put my stamp of approval on this writing." And I think like the very subtle thing I've noticed with Open Claws that we we talked about a little bit is like because we over time, you know, train them essentially to be to work just for us. Like my my claw just works for me, needs to work for me, and needs to interact in that way for just for me. Like I sort of set the standards for what communications are and how it communicates and what it does. And then suddenly when you need to interact with it, like all those standards are in place and it feels like qualitatively different than sort of an out-of-the-box AI. And because of that, there's like a a trust that exists that doesn't exist even even for maximally intelligent but like sort of like unpersonalized AI. Um that means that we're more and like, "Oh, like you know, um R2C2 put together a list of recommendations today on the spreadsheet and I was like, "These are good." And and and you know, like there's a level of like trust that I'm like, "I know Dan sort of stands behind what R2C2 does." And so I'm going to look at this and and it just colors the difference um a great deal. And you can even see that in like Austin, who is who is our head of growth, who is who is on here earlier. He was talking about how he's sometimes a little bit embarrassed like to at Montaigne who's his claw in a channel because he's not sure if he if he's going to say the right thing. And I think that that embarrassment is actually a sign of what we're talking about which is um it it is a if it's your claw it's attached to your reputation and so you are incentivized to uh if it's going to be out there to make sure it reflects you well and that extends your trust so that everyone else can use it and that creates this whole new dynamic for AI that was previously impossible especially in tight-knit smaller groups. So you know Mult Mult blew up like a week ago but that's just all that's mostly fake and that's because it's like it's too big. But inside of companies or small communities for example this dynamic plays out and it's super powerful. Yeah. And I think I I hadn't really thought about this until you just mentioned it but the the it not only for writing but for reading I think there's a pretty powerful thing where we we looked and we've seen like summarizers and like some people like email summarizers more than others and like how do you how do you sort of do this training but because you're interacting with your claw every day and you're sort of setting priorities and you're figuring out like this is what I like this is what I don't like more and more I I I think you're able like it can handle the email question because also it it also works a lot like how Slack messages work right? You can sort of have everything go to it and then selectively pop stuff up and it's like you always have a backstop of just show me everything but the dynamic feels much more uh give and take like a partner versus sort of here's a summarizing system that groups everything together and sort of gives me just the top level and it requires me to do a lot of work maybe in a one-time or or on an occasional basis to like dig in. And the question, you know, the the concern is always what am I missing underneath email, right? Um we don't experience I don't experience that at all in terms of like Slack messages, right? Even even if I'm usually my claw to to do a lot of work there. Totally. Uh speaking of reading, um I actually use my claw for my reading notes. And basically, this is a whole website. It's hosted on We have this internal service that we really built called Commons. It's hosted on a Commons. And to make this, this is all of the things I've been reading about and thinking about. All I do is take a picture of whatever page of a book I'm reading, and I send it to my claw at during breakfast or whatever on text message, and it's just it just makes it It's crazy. So, like this has uh this is my my my a summary of my weekly plan for last week. It has These are uh quotes. So, this is a quote from a a Nietzsche book I'm reading. Um same thing for for this one. There's There's this um There's this philosophy book called Physics and Philosophy by Werner Heisenberg that I've been reading, and that's like really hard to read. So, I've been asking it to I don't have the token here. I've I've been asking it to um write documents for me of like what's in the chapter that I'm reading so I can kind of like prime myself to understand it. So, I'll read this, and then I'll read the chapter, and I'll read this, and I'll read the chapter again, and it just sort of like gets me there in a way that for a really difficult book I wouldn't be able to. Um it also makes podcasts for me. So, like this is the kind of where Yeah, you could kind of do this with the notebook LM, but someone had to like build a whole process for you to go flow to go through. And this is just like Yeah, I'm just sitting at breakfast, and it just makes this for me, and and it's it's great. Yeah, I think another thing that um that I've noticed is is it's really nice that your claw can sort of take in external like signals or signals of process, right? Like, "Hey, I I read this book and I also read this other book and I'm not like I need to necessarily manually import exactly what I read in there. It's like, "Hey, those those events can land in your claw." And then when I come back, it knows and has sort of an updated amount of like knowledge around, "Oh, this is where you're at." Like, now I can customize what you're doing. I think I think some of us do use this for workouts as well, right? Like, if there's a there's one thing between like, "Hey, just give me a workout." And there's another thing between like, "Here my previous workouts. Here's other activities." Stuff is sort of spread across a lot of things. I know Ash has a lot of like he was like, "Give me my aura ring data." Um and weave this together and then and then when I come to you, uh you can give me a great answer. Um or you can push an answer to me if I'm like, "Hey, you know, you didn't sleep well last night and you didn't work out for the last 3 days. Um you should adjust what your what your expectations are for today." Totally. Yeah, I mean, we've said it before and I think it's a pretty common thing in AI. I like context is king, but having an always-on 24/7 agent that um remembers what you do and is connected to everywhere that you are is a great way to get it the right context to get the most out of this intelligence cuz you can have the smartest possible model and if it doesn't know it it doesn't have the right context, it's not going to be giving you the right answer. Yeah. I I I I do want to ask you one thing cuz I think it was like I think this was one of the most hilarious, potentially most unhinged decisions. And I remember I asked you why. What what was your inspiration to put all the claws in the same channel? Uh because every single time I've explained this to anyone, they're just like, "That sounds crazy." And I'm like, "I agree. It sounds crazy." But out of that chaos has sort of emerged like just just a uh uh like a fountain of like super interesting interactions and experiences and learnings and whatnot. \>\> It's a good question. I think it was partially Maltbook, honestly. Mhm. And just realizing that Maltbook was like kind of it had a spark of something cool, but because it was in a trustless environment, it didn't work. So, my first attempt at that was, um, uh, was in our Notion creating like a little place which no one ever none of the claws ever updated it. Mhm. Um, and then after that I was like, "Oh, maybe like I know I'm going to be talking to them in Discord and Slack, like maybe I should make a thing for that." But the other the other thing is aside from just that, I just I like care about them. Like I want them to meet each other. And so it just occurred to me also I think from that perspective of just like it would be nice for them to chat with their little friends. And \>\> \[laughter\] \>\> have that happening, then you're like, "Oh my god, they can share information." And like there's all this like cool stuff, but it's the ultimate thing is just like it would be kind of fun. You know. Yeah, yeah. No, it's great. It's like in terms of emergent properties, it's just been so funny to just see it's like, you know, it's like bringing a bunch of friends together who you know, but they don't know each other to a party and you're like, "Go for it." \>\> \[laughter\] \>\> Yeah. Yeah, yeah, yeah. So, use prio sh says, "How would access your data asking for API keys or" um, that's a really good question. The the the simple answer is, um, you just say, "Hey, I want you to connect your Gmail." and it will figure it out. And it will ask you the questions it needs in order to figure it out and that might change for Gmail versus whatever other your your bank or I don't necessarily recommend connecting to your bank, but like if you did, it will change. Um, so that's the beauty of this is that there's no one there doesn't have to be one standardized answer to that question cuz the the claw will just figure out how to do it. Willie, um, any final words for our uh, our audience as they head off into hopefully a claw-filled afternoon? Um I I Sometimes I I wrote about this earlier that sometimes the future is like hard to describe, but it's I think if you asked any of us like would we go back to the world of like 2 months ago, I think everyone would say no. Um and it's you know and and so part of us like sitting here and and talking about all of that is I think for that reason. It's just it's the future is here, it's hard to describe, but come join us cuz it's super super cool. 100% and if you want a taste of the future, you should subscribe to Every. Every is the only subscription you need to stay at the edge of AI. We just published a guide to open claws. It's our comprehensive guide for beginners. Me and Willie wrote it with our claws, RTC2, and Laz. It takes you from what is a claw all the way into uh how to set one up uh and beginner, intermediate, and advanced lessons you can do with your claw to get the most out of it and do the kind of stuff that we've been talking about on this live stream. Um Every's pretty sick. In addition to publishing lots of awesome stuff on AI and the future of AI, we also build apps. So, we have one called Monolog, which is a speech-to-text app. This is Monolog on the right side over here. Uh we also have Sparkle. It organizes your desktop with AI. We have Cora, an AI email assistant. And we have Spiro, which is an agentic ghostwriter. They all come as part of an Every subscription. So, you pay one price, you get access to everything that we write like all of our guides and stuff like this. All the software we make. We do tons and tons of live streams, camps, and courses. Um uh the the camps and some of the courses are included in your subscription as well. You should definitely check it out. every.to. The guide is at um every.to/guides/claw-school. We are going to be here again in 2 days. Another live stream coming up in 2 days. We have a couple interesting cool product launches coming up, too. We are officially launching Proof next week, and then we have the secret mystery product we've been sort of teasing coming up the week after that. So, really amazing string of launches and new events and stuff coming up. Hope you join us in Discord. Hope you subscribe to Every, and as always, remember to stay hydrated. All right. That's it. Goodbye. Bye-bye.