# AI Boardroom — Agentic Society Members · Clean Transcript
**Date:** 2026-06-26 · **Type:** AI Boardroom
**Location:** Industrious East 7th, 924 E 7th St (Conference Room A), Austin, TX
**Duration captured:** ~02:43:00 (in-room + remote)

> **Timestamp note:** times below are from the RAW recording (Fathom). Re-map them
> to the published/edited video before enabling any "watch at" links.

## Who was in the room
| Name | Business | Joined |
|---|---|---|
| Austin Distel | Agentic Society (host) · The Tox med-spas | in-room |
| Darby Rollins | Gen AI University / Agentic Society · SideHustle.live | in-room |
| Cristian | Agentic Society | in-room |
| Roman Bediner | Agentic Society (COO) | remote |
| AZ Moyer | AZ Does Taxes | remote |
| Prashant Vanka | Real estate development · 8 Homes AI | remote |
| Justin Day | Day by Day Productions (Local SEO) | in-room |
| Isaiah Zimmerman | Orectic.ai (OVAE platform) | in-room |
| Adam Weisberg | Restaurant business (first-time guest) | in-room |
| Jo Feathers | Wellness for autoimmune disease (co-owns an agency with Justin Day) | in-room |

---

## Transcript

### [00:01:16] Opening & room kickoff — Austin Distel
Austin gets the room going as people trickle in and the remote feed comes online. Prashant joins remotely, traveling and dialing in. Austin runs through the RSVPs — about 12 expected, with several still in transit or planning to catch the recording — then welcomes a first-time guest: "We have a new face in the room today — this is Adam." Adam says a quick hello.

### [00:12:10] How the boardroom runs — Austin Distel
Austin orients Adam and the room to the format: a few announcements up top, then Roses & Thorns — the round-table where members talk about the products they're building with AI — and finally Show & Tell demonstrations. He checks the remote audio and previews the day's final demo: Roman, the company's COO (dialing in remotely), will share a project-management Slack tool he built that keeps the whole team aligned on outstanding work. "As a team collective, it's been one of the most valuable parts of this month's build."

### [00:13:09] Announcement — the 7-Day Cloud Crash Course — Austin Distel & Darby Rollins
Austin opens announcements with the first thing the team shipped: Darby finished a **7-Day Cloud Crash Course**. He frames it inside three levels of AI development on the path to becoming an "agentic CEO." Level one — where most of the world still sits — is *chat*: you only use AI when you go to it, like a search engine. The course is built to move you and your team "from Cloud chat to Cloud code in seven days or less." By the end, everyone is caught up: they speak the team's language, have the tools and systems, understand how teams work and how to update the company brain, and they're building their own skills, connectors, and artifacts.

Darby explains the feel of it: "It's not a giant session of tutorials — it's day-by-day, brick-by-brick. Do this one thing, stack the skills on top, and by the end it all comes together. You just become more autonomous in how you work with it." Each lesson runs about 7–15 minutes with a progress bar across the seven days, so you can move through it and hand it to your team. The graduation moments include Claude Code / Claude design, but Darby says the most valuable part for him was learning how to *think* about the company brain — how people contribute to the company's collective intelligence. He adds that the jump from chat to Claude Code can feel intimidating, so Co-work is a solid stepping stone — for many people Co-work is more than enough — and once you understand how it operates, Claude Code just adds accelerant. Foundationally, "knowing how skills operate and running multiple skills on a cadence or schedule is a super low-hanging-fruit way to get stuff off your plate every day."

Austin agrees: "Most people use AI manually — it's only there when they go to it. It's not working in the background, not on a schedule." He points to Day 2 as where the rubber meets the road — turning a skill into a morning brief: "Almost every professional should have a morning brief." It's the introduction to *recurring* skills, the shift from AI to *agentic* AI. The whole course is about two hours of content and is live now in the Academy.

### [00:19:00] The Agentic OS thesis — AI is replacing headcount, not just tools — Austin Distel
At a high level, what the community is building together is the **Agentic OS**. Austin's framing: technology is no longer just replacing our tools and software — it's increasingly replacing *headcount*. "Not necessarily your current headcount, but your future headcount." Cost savings come down to two things that are separate: do you have the right intelligence, and do you have the right capabilities? "When you combine the ability to *think* with the ability to *do*, that becomes agentic — that's been the separator of 2026." In 2025, AI was mostly a thought leader sitting in your Slack; now, running on Claude Code or Codex, connected to your tools with API access, it can physically change your CRM or hit your point of sale — it takes on the actual availability of an employee.

So a skills library isn't just direction — "we're giving it capabilities, intelligence, and access to an org chart for handoff and collaboration." What the team has begun to build is a collection of **AI employees** you can install in your company. Austin uses becoming an agentic CMO as the example: you'd need a content marketer, a brand director, a media/show producer — roles that each cost real salary to hire (a content marketer might start around $60K/yr, a brand director around $100K, a media producer at least the cost of an agency retainer). Customer success, sales, chief of staff — all available as AI employees, "and you're going to get access to them today."

### [00:22:20] Live demo — the Skills Library & the librarian — Darby Rollins
Darby opens the library: a place to house and store skills, each mapped to where it fits in an org chart, each with its own description. You can search and browse, "but you can also just type in what you want, and the **librarian** goes and finds it — it pulls up the skills relevant to your search." He demos "hiring" an **Autonomous CEO Content Engine** skill: "This entire skill is copy-and-pasteable. It's also downloadable as an `.md` file. And it's callable through the **Agentic OS MCP** we've been developing."

He shows a worked example born from a Slack idea Austin posted — *is there a skill for transcribing videos and podcasts on Google Drive without downloading everything?* "I ran that straight through the pipeline we've been building, and we came out with a skill literally designed for that — it connects your Google API with a Gemini API so it can scan and transcribe any and all videos inside your Google account." You install it via a guided onboarding flow.

Then the third access path: "Everybody here has a Claude account — you can add this into the connector section of Claude right now." With the MCP connector you can say, "I want to talk with the Agentic OS — pull in my content marketer skill," whether you're in Claude, Claude Code, Codex, or Cursor. Every skill is also directly shareable with a teammate, or importable org-wide so the whole team gets it. Darby notes the skills get you 80–90% of the way there so you're not building from scratch, and there's a request mechanism — describe a skill you want, it runs through the team's production pipeline, and once ready it deploys into the library with a Slack announcement.

### [00:27:04] First impressions — the "ghost skills" problem — Member (Unattributed)
A member raised a thoughtful concern: "I love it — it's a great library and asset. One thing to be considerate about: we grab skills and implement them when we're in build-mode, solving a specific problem, and then a lot of times we end up with **ghost skills**. How do you keep up with all these skills and actually implement what you have — versus ending up with thousands of them in ten years? How do we not just build, build, build — but optimize, optimize, optimize?" Darby's answer: the librarian is a routing agent — clarity about the outcome you want lets it surface the two or three top skills you can chain together, and you can even build a "discovery skill" that recommends from your existing library before you add to the pile.

### [00:28:50] "Who, not how" — building roles, not just skills — Austin Distel
Austin shares his mental model: "The mindset I bring is *who, not how* — just like the book. Whenever I have to do something, I ask *who* would I hire to do this, not how do I do this." So instead of "create testimonials for the website," he thinks "I need to hire a customer marketer" — a real role he hired at Jasper, whose job includes reviewing transcripts and extracting testimonials. The team builds out *roles* with skills over time rather than a flat skill list; each role reads like a job description. His tip for building one: "If I were writing this job description to post on Indeed tomorrow, how would I write it? What qualities am I looking for? I'm giving them a soul. I'm giving them an identity."

He adds a Tim Ferriss move from *The 4-Hour Workweek*: to program top-1% expertise, look up the top books, YouTube channels, conferences, and communities in a field and bake them into the identity — "You're an expert copywriter. You've read Gary Halbert and others, and you apply those theories to your work." And then onboarding: "On another resource file you can say, 'onboard me into the company's knowledge around this — who are my partners on this project?'" Austin's punchline on where the leverage is: "The alpha is whether the CEO has the content they need to onboard the employees on their future roadmap. Do that one domino, and everything else gets easier. Your first week of onboarding used to take a week — I bet you can make that one hour when you onboard an AI employee."

### [00:32:50] Room discussion — building top-down roles & a content director — Members & Austin
A member named the click for them: "I've been developing skills and tasks that run throughout the day, instead of building out the *role* that has those skills attached. I've been building at level one when I should be defining the roles up top." Austin gives a live example of his own AI content director: "I just said, 'AI content director, recreate this style guide and build thirteen thumbnails for me.' Ten minutes later it came back with all the thumbnails, built perfectly to the style guide — but only because it had access to all the tools it needed." Once you frame it as building an *asset* rather than solving one small problem, it compounds.

### [00:34:20] Where it gets bigger — Claude Teams & a fully agentic team — Austin Distel
"This gets even better once the skills are available in **Claude Teams**, where it's no longer just the CEO who has access." Marketers on the team can communicate with the AI employees too, and it becomes a fully agentic team — collectively, Cristian, Darby, Roman, and Austin all share the same AI agents living in their Claude Teams account, some surfacing in Slack. Austin announces the rollout plan: the next seven days is **beta week** — "break this, play with it, tell us what you like and don't like — because next Friday I want to announce this to the public."

### [00:37:00] Bundled agent architectures, evals & the production pipeline — Justin Day & Darby Rollins
Justin offers his architecture model: rolling skills up into bundled **agent architectures** where a role-defined category sits over a subset of skills, so you can do a **hook invocation** on the agent that holds them — keywords, triggers, and webhooks fire the agent, and all its skills get invoked within the guardrails of its playbook. "That's exactly how my entire governance system is built — twelve agents, each with about six to eight skills, built for observability, tracking, and verification." A member asks how 170+ skills are quality-controlled. Darby explains the **skill production pipeline**: about seven or eight stages — a skill creator, a skill *architect* that reviews, then a series of "skill taste" agents that compare each skill against Anthropic's standards (YAML formatting, searchability), with the process improving as they learn. On portability: the skills are tuned for Claude but can be converted to Codex's `agent`-file schema via a converter (a `/goal` convert-all loop), targeted for next week so they work across models and tools like Conductor.

### [00:41:00] Access paths & the members portal — Austin Distel
Austin lays out how members will engage: **Option one** — go to the library, search, copy or download a skill as an `.md` file to your own device ("it's tangibly yours"). **The other way** — go to "connect," copy the link, and add the Agentic OS as a connector in Claude, then just talk to it: "What does the Agentic OS have for media buying?" The library will live as a page inside the **members portal** the team is custom-building. A member suggests a simple onboarding questionnaire (industry, business stage, pain points) that recommends starter skills so newcomers aren't overwhelmed by the size of the library — "like SaaS onboarding." Another flags that local downloads must be *completely* frictionless: the download should bundle an agent-handoff README so the agent auto-installs the skill as a plugin — "or they'll just end up with a bunch of dead `.md` files." Roman previews from the remote feed that there's another approach using a shared repository / Google Drive with "skinny" pointer links (symlinks) for the skills, which he'll show in his demo.

### [00:45:30] The new events page & "we made our website fully agentic" — Austin Distel & Cristian
Austin shows the new unified events page — every event type at Agentic Society on one page, filterable (e.g. the AI Boardroom), with upcoming events listed. It's a feature of the past two weeks' big build: "We made our website **fully agentic** — no one in the company is building this by pushing pixels. All of it is created with AI agents." This one is personal for him: after ten years on Webflow (speaking at their conferences, winning awards), he canceled his membership and hosting and moved Agentic Society over, recreating a site that originally took "two weeks and two days," now consistent against a synced style guide, posting to Vercel and GitHub. "I think this is the future of funnels, of website brand management, of SEO." Cristian: "That needs to be a clip — 'driving and vibing.'" Austin: "Today I took a 25-minute Waymo here and put up three website proposals on the car ride over, tethered to my phone."

### [00:49:05] Summer hackathon planning — Austin Distel, Prashant Vanka, Cristian & the room
Austin floats a summer workshop hackathon (date TBD) and walks three formats: (1) everyone builds the *same* agent together — he points to a screenshot of Darby running a Chief of Staff event with ~40 people all building the same AI Chief of Staff; (2) multiple speakers each teach how to build a specific part of a company with AI (SEO, company brain, Chief of Staff, marketing agent); (3) an open hackathon — two full days, ~8am–8pm, meals together, dedicated build time. "What we've discovered is you can get two weeks' worth of work done in two days." Ticket pricing for the two-day event would be set relatively low as a membership entry point (around $2,000).

Prashant proposes the format the room coalesces around — **pods**: "Start with a couple speakers to give inspiration and show what's possible, then break people into pods. Who wants to build what speaker A talked about? Speaker B? Speaker C? Beginners go build a company brain; advanced folks go do the harder thing. You get a piece of both parts — and someone paying for an advanced experience doesn't feel like they're not getting their money's worth." Austin builds on it: pods need multiple facilitators with expertise on each track, ideally the kickoff speakers also leading their pods so attendees always have "a sounding board to know they're going in the right direction — which is where a lot of people get stuck inside these hackathons."

Adam (the new guest) asks about skill level in the room. Austin sets a prerequisite: homework before arriving — most likely Darby's 7-day Claude Code crash course — "so you already have the right tools and connectors set up, you're not spending half the workshop getting your computer ready. After the crash course you're officially level two: you're using the terminal, you're not afraid of it, and you can speak the language of what a skill is and an artifact." Cristian endorses the multi-perspective, hands-on format from his experience in another mastermind — "it's so effective, it's a no-brainer." Prashant adds why protected time matters: "I walk away from a meeting with all these ideas to implement, and then I just get sucked back into work. Distraction-free build time is the whole point." Austin notes they'll add pre-event setup calls a few weeks out so attendees arrive configured, and a member suggests per-day milestone showcases so people see what got built and come back for more.

### [01:04:02] Roses & Thorns — the framing & Austin's share — Austin Distel
Austin frames the round-table: "This builds connection, helps us understand each other's businesses, and tries to make a breakthrough among the community — you're part of the collective intelligence of a lot of people." Roses = "what's blooming," time to celebrate; thorns = "what's painful," a chance to be vulnerable so you actually get help. He adds a prompt for everyone: *What's your vision for your company being agentic over the next 12 months — the dream state?* He also sets a mechanic: notepads on the table so anyone who can help a sharer jots their name down, with every rose and thorn captured in the next day's recap page for follow-up.

**Austin's rose:** "My rose is that the last two weeks has been so much building the thing I've been talking about with the team — Roman, Darby, Cristian. We put in a lot of hours creating something, and I feel really proud of the work. Just seeing what's possible." **His thorn:** his Dallas spa manager (The Tox) unexpectedly had to step away and leave the company, so he has to fly to Dallas next week to hire — something he's done many times, but it pulls against his vision. "My desire as an agentic CEO of The Tox is that I spend one day a week on it, and that I've enabled my whole team agentically to use Claude Teams to solve problems and onboard themselves." He reframes the thorn as an opportunity: "I've been thinking about what onboarding new employees looks like in an agentic company — getting them set up on Claude Teams as part of onboarding. My ask: what have been the most valuable parts of the first week of onboarding *your* employees — how do you activate and get them up to speed fast? If I can get my next studio manager going with AI, that will literally save a month of onboarding and dumb questions to me over text."

### [01:06:37] Roses & Thorns — Adam Weisberg (restaurant business, first-time guest)
"My name is Adam Weisberg. I'm in the restaurant business — about 26 years." **Rose:** his two-week AI learning journey — "creating a few dashboards: getting AI to work with OpenTable for a cover-count predictor so my people can prepare better 24 to 48 hours out, and a dashboard for daily and weekly P&Ls that communicates better with our financial software." **Thorn:** repositioning and re-marketing one of his restaurant concepts — "we're putting out different fish hooks to see what connects with our guests" — and underneath it, an industry that's behind on technology: "My industry is not technology driven. We're so far behind, and most of it doesn't talk to each other. My goal is to connect a bunch of these different pieces of software so my people aren't dipping into so many tools — it takes so much of our time, energy, and bandwidth." His upside vision: repackage that integration for other small restaurant groups, which he says is viable because the underlying tools are fairly concentrated and accessible via API. On the agentic vision prompt he was refreshingly candid: "I don't have a great vision for that yet — part of my journey is figuring out what's possible. There's a reason I'm probably the first restaurant guy in your group. I'm here in curiosity, and honestly to eventually build other businesses outside the restaurant world."

### [01:10:25] Roses & Thorns — Darby Rollins (Gen AI University / SideHustle.live)
"My name is Darby Rollins." **Roses:** shipping the MCP and skills library the room just saw, and rebuilding the website for his comedy game show, **SideHustle.live** — "Shark Tank meets Whose Line Is It Anyway" for entrepreneurs — with Claude, hosted on Vercel. "The old site sucked at articulating what the show is about; I spent a lot of time with a Claude project on it and I'm really happy with how it looks — and it doesn't just look like an AI site." His 12-month vision ties to the show: "Get collectively millions of views on our content, then go on a world tour in 2027. I want to build the ultimate show/event production and booking agent — that books events, captures the content, has the editor complete the full episode, and runs it through an entire distribution system, without me manually scheduling everything." **Thorn:** a separate venture with a real-estate partner — an outbound lead system (targeting Texas real-estate agents selling land, building customized outreach, scraping via APIs since MLS blocks scraping, emailing through Instantly). "It's been a much slower process than we wanted — it's a lot of technical configuration right now. There's no ask, it's just a thorn — but I'm trying to get it where I don't have to be the person running each stage." Austin connects him with Prashant on the spot; Prashant offers to compare notes on the real-estate model and they agree to connect.

### [01:19:07] Roses & Thorns — Prashant Vanka (real estate development · 8 Homes AI, remote)
**Roses:** "Lots of cool stuff in our real-estate development world — we're building 56,000 square feet, and debt and equity are getting easier to raise. The fun question is whether to overbuild for Austin again or go to a second market — and do I add a second asset class or open a second market?" He's also an investor in a restaurant group expanding to several major markets over the next couple of years. **Thorn:** his AI real-estate acquisition agent (8 Homes AI), which goes direct to sellers on distressed deals. "I got it doing automatic texting and qualifying for intent, motivation, ballpark price, and urgency — but after that, everything breaks down. I don't think my agent harness was ever organized the right way, so it degrades over time. I need to organize the second brain correctly, put everything into skill files, build them into workflows so it doesn't deviate, hallucinate, or get worse." He ties the fix to scale: get objection handling, underwriting, and offers consistent, then expand across multiple markets — a clear path to significant monthly revenue per market if the systems hold. Austin offers a concrete next step: a team-built **company-brain architecture playbook + two-hour training series** (modeled on Darby's Claude Code series) within ~30 days. Prashant: "That'd be awesome."

### [01:23:37] Roses & Thorns — AZ Moyer (AZ Does Taxes, remote)
"I'm AZ, I do taxes here in Austin." **Rose:** "We've been building skills in Claude that take our brand book and reference it every single time, so every deliverable in our entire ecosystem is on brand. Distel was one of the first people who taught me about good design, so it's always been important to me — but our designer isn't on staff every day. Now, funny enough, five times in the last two weeks clients have signed and then said, 'Wow, that's really well done, that's impressive' — and I'm like, 'Claude helped me out.'" It's translated into meaningful new pipeline in just two weeks. **Thorn:** "There's a lot of busy work in our world — research, calculations, spreadsheet compilation. My real value is the *delivery* of data analysis: talking to clients and showing how their financials and transactions affect their situation. The more we take busy work out, the more we can scale and the more people I can help." He adds a confidence angle: "One of the first sessions was about imposter syndrome — people say 'AZ, you should do the math at the restaurant,' and I'm like, I'm an accountant, not a mathematician. This has given us the authority and confidence we need." Austin's nudge: spend time on Claude Design, not just a static skill file. AZ: "Definitely need to spend more time on that piece."

### [01:36:00] Show & Tell — the internal PM agent fleet — Roman Bediner (COO, remote)
Roman, with a hard stop, jumps the order to demo. Two caveats up front: "What we built here is internal for our team — what Darby's been showing with Agentic OS is member-facing," and "there are multiple ways to skin a cat; just because I did it a certain way doesn't mean you have to — hopefully it sparks ideas." He confirms the room lives in Slack, then triggers three agents from Claude to introduce themselves live.

**Slack PM** — the flagship. "Every morning it pastes a virtual standup, because we don't need another call on our books. It has a little personality: 'Last standup of the week — reply in thread: what are you focused on today, what did you complete, what's blocked.' Then it breaks the day down per person — 'Hey Roman, this is what you've got today; Darby, that's your stuff; Distel, that's yours.'" He shows the loop live: he'd replied that morning that he was demoing Slack PM at the mastermind, and the next cron run replied back in character. The data lives in a **Slack List that exposes an API**, surfaced as a **Slack Canvas Kanban board** grouped by owner, deliberately split into three buckets: **Action Items**, **Open Loops** ("we mentioned it once but never discussed it"), and **Decisions Needed." His one-line data flow: "List → API → Slack thread."

**Chief of Staff** — "the team's intelligence layer. I turn raw inputs into structured knowledge the whole team can act on." It processes mastermind transcripts and Loom links, pulls live pipeline data from GoHighLevel, watches Slack for signals, and writes a Monday "week ahead" / Friday "week in review" digest — and deliberately doesn't run the tracker so it doesn't step on Slack PM's toes. Austin narrates the always-on behavior: "Anything we say in Slack in natural language, these guys are watching and organizing. I just told Roman I'd send him a GitHub of my Airbnb microsite — because I said it out loud and it was listening, after this transcript is processed it'll put a task on the board assigned to me, keeping me accountable."

**Agent Orchestration Director** — the newest, "the operations layer that runs the fleet. Think of me as a manager of agents — I keep them healthy, coherent, and coordinated. I'm the fleet's memory so we stop re-solving the same problems." It monitors the fleet, is the single alerting layer if an agent fails (it DMs Roman), routes work between agents so a Chief-of-Staff action item lands on the PM tracker, and owns deploys, cron schedules, and fleet memory.

**Architecture.** The conceptual core is the **skinny skill** — a pointer/symlink. "The only thing Claude knows is: you're the Slack PM for Agentic Society; before doing anything else, use the Google Drive MCP to read your instructions from the company-brain skill library." That library on Google Drive holds the real source of truth for each "employee" — changelog, inputs/outputs, personality, READMEs, test cases, workflows. "Now it's a shared skill — it doesn't just live on my local Claude. If Darby needs a new capability, he tells Claude Code to add it, it updates the Google Drive, and everybody on the team takes advantage of it." The stack: **Vercel** runs the cron jobs and a JavaScript handler that runs the agent loop in the cloud; **GitHub** is the code source of truth ("the bodies"); **the brain lives on Google Drive, not the repo"; **Slack API** is the interface (secrets stored on Vercel); **GoHighLevel** is integrated so Chief of Staff can pull the CRM pipeline. Roman is emphatic: "I'm not a software engineer. I've never written a single line of code in my life. All of this is done by Claude Code, including the documentation — it draws the architecture diagrams and creates the whole codebase. Nothing lives on my local Mac — that's a condition for every skill I build, because we all have a laptop, a Mac Mini, a phone."

**The Q&A.** On cost: "All three agents cost us about **$15 a week** in API usage — about five bucks per agent — because once I implemented **prompt caching**, the cron jobs got cheap." Model: **Sonnet 4.6** is the default for the agents; Roman uses Opus 4.8 personally for reasoning and design. On reliability: early retrieval issues went away once the Orchestrator added observability — "if a cron job fails, it DMs me, and now I'm training it to self-heal: defensive code, retries, backups." A member recommends Sentry (free via the Vercel marketplace) for traceability; Roman takes it as a next step. On scale: deploying the third agent took about a day versus a couple of weeks for the first, because the infrastructure already existed — "from there it's just training, training, training: more capabilities, more access to your CRM, accounting, member profiles."

Austin's reaction frames the economics: "A project-manager role can start around six figures a year — and this works better than that for keeping us aligned. How much stuff just gets lost in Slack? We ran into exactly that less than two days ago." And the org-design thesis: "Organizations will become more flat — more directors and executives than individual contributors. The future hires at agentic companies are people who maintain and manage their own AI direct reports. It's Roman's job as COO to manage his — the Orchestration Director, then the PMs and Chief of Staff." He stresses it was built by non-devs: "Roman and I are equal in our understanding of technology — I can barely edit HTML." Darby and Cristian, who use it daily, vouch: "You forget you said you'd do something, and it reminds you every day. It's so useful — I love that it nudges you." Roman commits to sharing the architecture diagrams, signs off to his hard stop, and the room returns to Roses & Thorns.

### [02:10:51] Roses & Thorns — Jo Feathers (wellness for autoimmune disease)
**Rose:** a clarity breakthrough — her business isn't coaching, it's an AI-powered wellness ecosystem. "I'm not selling a fix for any one condition — I'm selling **behavior change**, which is the concrete base for the majority of things you want to improve. Anything you want to improve comes down to our behaviors and habits." The wellness space is flooded with noise about what's "best," so her value is education that cuts through it — showing people what matters most and what their actual first step is. **Thorn:** "I have my wellness assessment built out in Lovable and ready to go. Great — what if I push that out? I have no funnel, nothing to push these people into." Her agentic vision is a stack: an intake-specialist agent that runs the assessment (sleep, habits, nutrition), then pattern-recognition agents and educational agents — possibly housed in Skool — "but I don't want it to be just AI; people still need accountability and real-world support. There's only one of me." The room riffs on a branded "Jo Bot." Austin shares the model of his wife's doula, who built a custom agent in her own voice trained on 25 years of practice to field first-time-parent questions — "designed not to replace the mother's intuition but to ask questions and co-create a result." Austin offers to help Jo reconstruct her frameworks from old Zoom-cloud transcripts and Word-doc notes, and points her at the now-live 7-day onboarding. **Ask:** build the knowledge-base bot that houses her brain and automate her systems so support feels organic.

### [02:18:09] Roses & Thorns — Justin Day (Day by Day Productions, Local SEO)
**Thorn first:** "My team's been crushing it and finally pulling me out of the day-to-day, to the point where I haven't known what life is like — I can sit and build. But that's the thorn: I start one small thing and suddenly I'm building something massive I didn't intend to. I need to stay in my lane on what drives revenue right now." **Roses:** it's his birthday, and he chose to spend it in the room; and he closed a solid new client deal that morning ("the 8 a.m. closer"). The business is going well, and he's launching a new front-end offer at the end of the week. He explains the pivot: his original front end, the Invisible Business Scan, was so all-encompassing it overwhelmed people; his Facebook ad data showed the winning creatives (strong click-through and opt-in rates) were all built around Siri. So he thinned it to a single "Siri score." The timing thesis: "Apple just announced Siri runs on Gemini, so your Google Business Profile is more important than ever." The new offer is **$97 to fix your Google Business Profile** and schema — branded "Siri Superpowers" — upselling into an agent-readiness website review. "Homeowners with money have iPhones that use Siri — the more ingrained I can get, the better off I'll be." **Ask:** keep the conversations, accountability, Slack, and calls going — "the best support I could possibly have."

### [02:22:07] Roses & Thorns — Isaiah Zimmerman (Orectic.ai · OVAE)
"My name is Isaiah Zimmerman — I've been a builder in the AI space for over four years." **Rose:** meeting a likely advisor, **Neil Conran** (a Marine Corps veteran, ~20 years in tech), at a Conscious Builders dinner — they "jammed for two hours." Neil had backed a well-funded company doing essentially what Isaiah does, but for enterprise. The key insight: "What changed their positioning was reframing as an **AI lab** — AI labs are more agnostic about the products they deliver, and that's what unlocked the enterprise category." **Thorn:** positioning and differentiation. "I inadvertently built an enterprise-grade GraphRAG infrastructure — a company brain with managed AI and observability — and it's too much for small and medium businesses. Not from complexity; it's just not needed at that size." So Orectic has become a methodology and philosophy, and OVAE is the delivery layer. He resists verticalizing — "everyone says niche down, go solve this for HVAC; I don't want to, because the platform can do anything so well" — but concedes proving the methodology in a few service-adjacent verticals could make the enterprise product more valuable. He lays the pain bare with his own sales-call script ("What do you do? We build managed AI infrastructure. What the heck is that?...") and diagnoses himself: "I'm selling on mechanism because I'm technical — I'm not selling on outcomes. Almost every deal dies at the same place. That's the market signal." His other rose is grounding — kids thriving, wife happy, everyone healthy. He notes he's built 50+ active systems in five months (15–20 fully autonomous), the newest a Fathom-webhook "offer-market feedback loop" (nine phases) that acts as a private AI sales coach. **Ask:** help refining the positioning problem; Austin offers support and a call.

### [02:28:57] Roses & Thorns — Cristian (Agentic Society)
**Rose:** "I've been loving building closely with Austin — GoHighLevel, Zapier, Claude, the company brain, onboarding, connecting everything." He landed his first client this month and moved to a new place — "a big month." **Thorn:** prioritization, specifically protecting time for Agentic Society's growth — inviting people to events and moving them down the funnel into the sales pipeline. "That's the biggest thorn — making sure to take care of that rock." He's also built a personal content engine over the past two weeks: a marketing agent that researches formats that work, plus a pipeline that turns coaching calls into carousels and scripts and clips long content into multiple distributable pieces — "all within Claude." **Ask:** hold and execute the dedicated Agentic Society growth/pipeline block already scheduled with Austin.

### [02:30:35] Shout-out — Cristian's viral video & the content lesson — Austin Distel & Cristian
The room celebrates a clip of Cristian's that hit roughly a million views in a single day across TikTok and Instagram — an AI-music bit that earned him a genuine Suno sponsorship. "We filmed that in less than ten minutes." Austin pulls out the lesson: "When you find a winning format, do it again and milk it — and work even harder while it's working. Most people get a winning format and immediately try something else." He notes Cristian grew from ~2,000 followers at the start of the year to ~140,000 through relentless testing. The room adds the current content meta: combine two trending things in your own way, use pattern-interrupts (a review from a car, a walk-and-talk, an unexpected visual) to beat the "dopamine drain" and hold attention.

### [02:12:14] The post-event recap page (product walkthrough) — Austin Distel
Austin demos the capability rolled out the prior week: after every boardroom or mastermind, the website agentically generates a full recap page — the video replay, a downloadable transcript members can drop into their own Claude, a room photo, the three biggest "frame-breaking" topics, supporting links, and key quotes from attendees (he cites Rohan's line from the prior week: *"In a world of abundant information and knowledge, the only thing that truly is scarce is a trusted relationship."*). Everything is shareable to X or LinkedIn straight from the page. Crucially, the page extracts each person's Roses & Thorns *and* their **ask**, so commitments ("I'll give Isaiah a call on positioning") are tracked, and members who want to connect get the relevant links. Austin floats a social account built purely from in-room quotes; Cristian volunteers to own it. "With every one of these boardrooms, we're getting better."

### [02:16:35] How the recap is built — the three-recorder transcription method — Austin Distel & Isaiah Zimmerman
Austin shares the methodology behind transcript accuracy: he runs three recorders — a Plaud pod on the back of his phone, Fathom, and Granola — and had AI score how each performs. Each has tradeoffs (Granola is good at separating speaker turns; Fathom enriches the direct transcript with more color), so the verified "trifecta" is pairing two or three together into one enriched transcript, because cross-checking catches errors (one tool spelled "Darby" as "Derby," corrected against the Zoom recording of who was present). Names are further verified against the calendar RSVPs, and each member's profile is enriched meeting-over-meeting so it hardens over time. Isaiah corroborates with his own internal tool, **Room Oracle** (built on Deepgram), which diarizes in-room speakers using the Mac's three microphones and streams notable ideas and questions in real time — a two-hour session can yield ~150 questions; he once let it run two days and got ~1,300, which he fed back to his company brain to audit its retrieval.

### [02:16:04] Close & next steps — Austin Distel
Austin wraps: "That wraps up today's boardroom — thanks for coming." Adam, the first-time guest, thanks the room ("thanks for letting me crash the party") and the group takes a photo. Austin moves straight into filling the next event, which kicks off seven days out: a CRM-wide registration email goes out today (Austin will send it this time and record a Loom so Cristian can run it next time), and Cristian will load his compiled prospect list into GoHighLevel and work it with personalized text-message invites.

---

*Captured and enriched from the in-room (Plaud) and remote (Fathom, Zoom) recordings, cross-verified against the calendar roster. Timestamps are from the raw recording and should be re-mapped to the edited replay before use.*