📖I'm About to Build an Entire Agency Live This Weekend. Solo. Unedited. Probably Sweating.

Here's a thing that's been eating at me for months.

I keep watching self-proclaimed "agency gurus" promise to teach people how to build a GoHighLevel agency... and then they never actually build one. They show you a slide deck. A theory. A vision board. Maybe a redacted screenshot. They'll talk about snapshots for 40 minutes without ever opening a single real account. They'll teach you how to sell services they've never delivered. They'll charge you $2,000 to watch them point at a Loom recording of someone else's work.

It drives me crazy. Because the people paying for those courses — they're trying to change their lives. They're trying to escape a 9-to-5. They're trying to build something real. And what they're getting is theater.

I've built thousands of these. Real GHL agency accounts. Real sub-accounts. Real snapshots — onboarding workflows, calendar setups, missed-call text-backs, AI receptionists, voice agents, payment integrations, white-label everything. I've made every mistake you can make in this platform, fixed it, then made the next one. That's how you actually learn this.

So I'm doing something nobody else is doing.

This weekend, I'm going to build a complete agency. From zero. Live. Solo. Unedited. Two days. You'll watch me set up the agency account, configure the white-label, build out the first sub-account for a client, create the snapshots, wire up professional email, configure calendars, layer in the voice AI and chat AI, walk you through which services to sell, and show you exactly how to sell them.

Not a slide deck. Not a faceless tutorial. An actual real-time build. Mistakes included.

I'm calling it the Build-A-Thon. And I'm calling it that because I'm not going to edit the screw-ups out. You're going to see me hit errors. Watch me recover from them. Hear me talk through what I'm thinking when something breaks at the worst possible moment. THAT's the part nobody else shows you — because they don't have the reps to recover from a real error in real time.

Pray for me that I don't make too many mistakes. Some are going to happen. That's fine. Honestly? Some of you will learn more from watching me debug live than from any "perfect" tutorial you've ever bought.

This is for you if:

• You're an agency owner currently struggling to build out client accounts cleanly or sell your white label

• You're working a 9-to-5 and dreaming of a part-time agency that eventually replaces your day job

• You're ready to change careers and you've been waiting for someone to actually SHOW you the work, not just sell you the dream

And here's the part I want to be crystal clear about: it's free. No catches. No "free, but you have to sit through a 90-minute pitch." No upsell hostage situation. Free as in free. Register, show up, watch me build, take notes, ask questions in the chat.

There will be bonuses for people who show up live. Special guest speakers dropping in with discounts on tools I actually use in my own agency. Live giveaways. The whole event is engineered to deliver a lifetime of education in two days.

Register here for free → https://live.automatedmarketer.net/buildathon

If after the weekend you want to keep going deeper, I'd love to have you inside the Skool Community — that's where the post-event Q&A, the SOPs, the recordings, and the daily back-and-forth happens. Free to join: https://www.skool.com/automatedmarketer/about. And if you want VIP-level access — direct hot-seat support, advanced templates, the inner circle — that's $47/mo at https://www.skool.com/automatedmarketer/plans.

Here's the bigger truth I want you to walk away with this weekend, whether you make it live or not:

Building in public is scary. Building solo on a stage, while thousands of people watch, with no edits, no safety net, and no "let me cut to a pre-recorded segment" — that's terrifying. I know I'm going to mess things up. I know somebody's going to screenshot a typo. I know I'm going to forget to flip a toggle and have to come back to it 20 minutes later wondering why my email isn't sending.

I'm doing it anyway. Because the only way to teach people how to actually build something real is to actually build it in front of them. The polished version isn't education. The polished version is marketing.

So this weekend, you get the real version.

Show up. Bring questions. Watch me sweat. Take everything I give you. And then go build your own.

Wish me luck.

— Nuno

🚀 The Top 1% of Closers Aren't Better Than You. They Just Stopped Guessing Who to Call Next.

The best closers I work with aren't outselling your reps. They've just rigged their day so they never have to decide who to call next. Every dial is the highest-priority person in the queue — automatically. No thinking. No guessing. No wasted energy. The video below is the first public demo of the feature that makes this possible inside HighLevel. Watch it. Then come back so we can talk about how to actually use it.

Open your CRM right now. I'll wait.

How many leads are sitting in there that nobody's touched in over a week? How many came in from your weekend ads that nobody called back Monday morning? How many people filled out your form, didn't hear from you for 3 days, then went and signed with someone else who actually picked up the phone?

Don't answer out loud. Just sit with that number for a second.

Here's the thing — every one of those uncalled leads is a $50 bill you set on fire and walked away from. (Don't worry. I've done it too. We all have.) The painful part isn't that we're bad at sales. It's that we're great at lead gen and bad at the part that actually closes the deal — picking up the damn phone, fast, and knowing exactly who to call next.

That's the gap Wave is built to close. And it's exactly why the closers in the top 1% don't look anything like the closers in the bottom 50%.

Wave is already the #1 downloaded app in the entire HighLevel marketplace — and there's a reason for that. Fully embedded as a sidebar in your CRM. No platform switching. No copy-pasting numbers. Multi-line dialer that calls 3 people at once so whoever picks up first is the conversation you're having. Dispositions logged. Recordings stored. Transcripts saved. All inside your existing CRM. No bolt-ons.

But this newsletter is about what they just shipped: Call Boards. First public demo. This is the piece that takes you from "we have a dialer" to "we have a sales operation that runs itself."

Here's what Call Boards actually does:

• It builds a prospecting pipeline OUTSIDE your main CRM pipeline. Your sales pipeline stays clean — warm leads, appointments, closes. Your prospecting pipeline lives inside Call Boards — new, working, follow-up, dead. No more clogging up your actual deal flow with 400 cold-call statuses.

• One button. The right person, next. Your rep (or you) hits "Call" and Wave loads the queue in priority order — overdue callbacks first, then hot leads, then new, then working. They stop thinking. They just dial. The mental tax of "who am I supposed to call right now?" disappears. This is the 1% move.

• Hot lead detection from your forms and ads. A new lead hits your funnel — Wave auto-bubbles them to the top of the queue with a 🔥 emoji. Your speed-to-lead drops from "whenever someone gets around to it" to "the next call in the queue." That's the difference between catching them while they're still researching and calling them after they already signed with your competitor.

• Scheduled callbacks that don't slip. Mark a callback for Thursday at 4 PM — at 3:58 PM Thursday, it pops into your queue, pauses whatever campaign you're running, surfaces the contact. No sticky notes. No calendar reminders you snooze 4 times. Just the right person at the right minute.

• Dispositions trigger HighLevel automations. This is the unlock most people miss. Mark a call "interested but needs follow-up" — fire a 7-day nurture sequence. Mark "no answer" — trigger a text+email cadence. Mark "callback scheduled" — send a confirmation SMS. The disposition isn't a note. It's the trigger. Your team logs the call once. The automation handles the next 14 touches.

Real talk: I need you to read this next part twice. This is not just a tool for your clients. It's for YOU.

The same agency owners running flawless GHL builds for their clients have a graveyard of unfollowed-up leads in their own CRM. We get so good at fixing everyone else's bottleneck we forget to look at our own. The cobbler's kids have no shoes. The marketer never markets to himself.

Install this on your own account first. Run your own leads through it for two weeks. Then sell the service to your clients with the confidence of someone who actually uses the thing.

“🔥 Pro Tip: Use Wave on yourself first. The dual play most agencies miss — install Call Boards on your own business, dial your own neglected leads for 14 days, and document what happens. New revenue. Recovered deals. Booked calls from leads you'd written off. Then turn that into a productized offer: "Sales Acceleration Setup — I install Wave, configure Call Boards for your exact sales process, wire up 3 disposition-triggered automations, train your team in 60 minutes. $1,500 setup, results in 7 days." You're selling the offer with proof from your own business — not a vendor demo. Then stack Wave's preferred partner program on top — referral commissions auto-pay every time your clients use your promo code. You get the setup fee. You get the MRR commission. You get the case study. All from a tool you were going to use anyway. “ —

One more piece worth watching: Wave's spam protection and phone number remediation feature is in beta and shipping soon. If you've ever had numbers go cold or get flagged after a few thousand dials, that's the upgrade. Most teams find out their numbers are spam-flagged the same week their conversion rate cratered. This kills that blind spot before it costs you.

Here's where this all lands. Your leads are not the problem. Your offer is not the problem. The chaos between "a lead came in" and "someone called them at the right moment" — that's the problem. That's the leak that's been quietly bleeding revenue out of your business while you've been busy building funnels for everyone else.

The top 1% of closers aren't lucky. They aren't gifted. They've just removed the one decision that was killing their output, and now they make 3x the dials at twice the conversion rate of everyone else in their company.

You can do the same thing this week. Or you can keep guessing.

Plug the leak. Not next quarter.

🎯 Get 10% off your first month with my code. Sign up here: https://refer.wavv.com/?fpr=wuk2w — and drop the promo code NEW-WUK2W at checkout. That's a free taste of what happens when you stop letting hot leads die in your inbox.

 🤖 Claude's Been Guessing About Your Business This Whole Time. Build It a Real Second Brain in 20 Minutes (Free).

Here's the dirty secret about Claude Code, Cowork, all of it — most of the time, it's not pulling from your actual setup. It's pulling from its training data and educated guesses. And you don't notice because the answers SOUND right. The video below walks through a free, 20-minute build that gives Claude a real second brain it can actually reason against — not just trust-fall into.

You've asked Claude a question about your business and gotten a confident, beautifully written answer that was completely wrong. Don't pretend you haven't.

Here's the thing — Claude isn't lying to you. It just doesn't actually have your information. So it predicts what the answer probably is. And prediction sounds like knowledge. That's the trap most people fall into and never climb out of.

If you've ever fed Claude a 14-page SOP and watched it confidently butcher the answer two prompts later — that's why. Memory in Claude products is great for "Nuno likes navy blue and runs an agency." It's terrible at "the exact 7 steps Nuno uses to onboard a new GHL client, including the part where the setup fee gets collected BEFORE provisioning the snapshot."

For real business knowledge, you need a real brain. Not memory. Not vibes. A structured, graph-connected wiki Claude can actually read from.

That's what we're building. Today. In about 20 minutes. For free.

The stack: Obsidian + Claude Code. Obsidian is a free note-taking app that runs locally on Mac or Windows. Claude Code does the heavy lifting — it builds your wiki structure, organizes your raw materials, creates the cross-links, and updates everything when you drop new files in. You're not building anything by hand. You're feeding it raw materials and watching it construct a domain expert in real time.

Here's what this actually unlocks:

• A living wiki built FROM your raw files. Drop in transcripts, screenshots, web clippings (there's a free Obsidian Web Clipper Chrome extension that clips websites straight to your vault), PDFs, anything — Claude auto-organizes them into a graph of interconnected wiki pages. You watch the graph nodes appear in real time as it builds.

• Connections you wouldn't make manually. Claude auto-creates [[wiki links]] between related concepts. Ask about Topic X later and it pulls in everything connected — not by keyword match, but by the relationships it built.

• Reasoning over STRUCTURE, not just semantic similarity. This is the key difference from vector databases like Pinecone. Claude follows the graph the way you would mentally — schema first, then index, then the specific pages, then the links between them. It's how a smart consultant thinks about your business, not how a search engine thinks.

• A "hot cache" file that makes every future query 10x faster. Claude generates a synthesized summary of your entire wiki as a single markdown file. Any future agent, skill, or chat loads that first — instant context, no token bleed.

• One brain per domain. Build one for a tool you're learning (the demo in the video builds one for Paperclip). Build one for your agency operations. Build one per client. Every brain is a vault, every vault is its own expert.

• Free and local. No SaaS subscription. No monthly fee. Runs on your machine. Your data never leaves unless you choose to push it somewhere.

Real talk on the limits — because nobody else is telling you this part:

• This isn't a dump for your entire YouTube channel. Keep it under ~100 sources. Past that, the index and schema start eating tokens and the magic fades. Curated beats bulk every time.

• Name your pages clearly. This setup finds by page name, not by semantic meaning. Sloppy naming = Claude can't find things.

• Re-drop files when you update them externally. No auto-sync. You change the source, you need to re-feed it.

• For massive archives, marry this with Pinecone or Supabase. The Obsidian brain handles structure and reasoning. Pinecone handles scale. You can run both. Future video on that one.

Grab the full step-by-step SOP here: https://docs.google.com/document/d/1LVElPJ2QZwijApBUUTD8WDF1gTjUHokb/edit — every command, every prompt, every folder name. Copy, paste, build. Done in an afternoon.

🔥 Pro Tip: The agency play here is enormous and almost nobody is doing it yet. Stop selling clients on funnels, ads, and chatbots — those are commodities now. Start selling them an AI Infrastructure Build: a custom Obsidian + Claude Code second brain trained on THEIR business — their SOPs, their scripts, their vendors, their tribal knowledge. Charge $2,500 to $5,000 for the initial brain build, then $300 to $500/month maintenance to drop in new files, run quarterly wiki lints, and regenerate the hot cache. Position yourself as the AI Infrastructure Consultant for your niche — real estate, dental, plumbing, whatever you already know. Three retainers at $400/mo is $1,200 MRR for roughly 90 minutes of work per client per month. And the moat is real: once you've built their brain, every other agent or automation they ever want lives on top of YOUR foundation. They can't easily leave. “ —

Here's the bigger pattern. The people building actual long-term AI leverage in 2026 aren't the ones with the best prompts. They're the ones treating their knowledge as infrastructure. They build it once, structure it right, and every future agent, skill, and workflow runs on top of that foundation.

Everyone else is going to keep pasting files into chat sessions, getting confidently wrong answers, and wondering why their AI feels generic.

Build the brain. This week. Then watch what becomes possible on top of it.

🧠 Voice Agents Got Real. Anthropic Went Full Send. And Elon's Now Selling Compute to His Biggest Rival.

This was one of those weeks where you blink and miss four things. OpenAI quietly dropped the voice agent infrastructure that's been sitting in everyone's roadmap for two years. Anthropic raised usage limits while signing compute deals with SpaceX (yes, that SpaceX). And we got the first real look at where AI agents are headed next. Buckle in.

OpenAI Built the Voice Agent Stack You've Been Waiting For

Three new realtime voice models shipped from OpenAI this week. The implications for anyone running a GHL voice agent, AI receptionist, or sales callbot are massive.

• GPT Realtime 2 — uses GPT-5 class reasoning during voice conversations. Your AI can actually think through complex requests in real time, not just pattern-match.

• GPT Realtime Translate — live speech translation from 70 input languages into 13 output languages, KEEPING PACE with the speaker. Listens for the verb, starts translating mid-sentence, handles interruptions in different languages without losing the thread.

• GPT Realtime Whisper — streaming speech-to-text live as the speaker is still talking.

• The killer feature for agencies: voice agents can now stay quiet in the conversation. In one demo, the AI listened while two humans talked, didn't interrupt, only re-engaged when called by name. The difference between "creepy AI in the call" and "useful AI listening on the call."

• Currently API-only. The moment this hits Cowork and Codex (probably within weeks), every GHL voice agent build levels up.

Anthropic Went Full Send (And Elon Is Suddenly Helping)

Anthropic shipped four major updates this week. Here's what matters:

• Higher usage limits across Claude Code and the Claude API. The single most-requested fix from heavy users. Hitting limits at 2 AM should now be a memory.

• SpaceX compute deal. Yes, that SpaceX. Elon Musk was publicly trashing Anthropic as recently as March. Now he's selling them compute. The read: enemy of my enemy. Elon's in lawsuits with Sam Altman. If Anthropic pulls ahead of OpenAI, Elon wins.

• $200B Google Cloud commitment on top of the SpaceX deal. Anthropic is loading up for a sustained Code/API push.

• Claude embedded across Excel, PowerPoint, Word, and Outlook — with memory persisting across all four apps. Thread in Outlook, jump to Word, reference what you were just discussing. This is the enterprise lock-in play. Microsoft is about to have an AI integration story competitors can't match for a year.

The Dreaming Update — Where AI Agents Are Actually Headed

The most under-the-radar Anthropic drop this week is the one most likely to change how you build with Claude in 2026: a feature called Dreaming inside managed agents.

• What it does: Agents review past sessions on a schedule, find recurring patterns, surface mistakes you keep making, restructure their own memory to stay high-signal.

• Not ChatGPT-style memory ("Nuno likes navy blue"). This is proactive agent self-improvement running in the background.

• The direction this confirms: AI is moving from reactive ("ask a question, get an answer") to proactive ("I noticed this pattern, want me to fix it?"). OpenClaw and Hermes have been pushing this for months. Now Anthropic is in.

• Status: Only in managed agents right now. But it's coming to Cowork. That's a near-certainty.

Rapid fire — the rest that matters for your business:

• GPT 5.5 Instant is now the default ChatGPT model for everyone (free AND paid). More concise, more personalized, baked into Microsoft 365 Copilot. Worth re-running any prompts that depended on the old verbose output.

• Codex + Chrome plugin is live — Codex now drives your Chrome browser directly. Still buggy on day one. Watch this for next-gen scraping and research workflows.

• Grok 4.3 dropped — solid model, dramatically cheaper than Opus or GPT-5, not state-of-the-art. Good fit for high-volume, lower-stakes workloads (bulk content classification, mass tagging).

• Sam Altman & Mira Murati testimony is the gossip of the week. The text messages from the day Sam got fired leaked publicly. Wild weekend read if you want to see how OpenAI runs internally.

• Adobe Acrobat AI — chat with PDFs, generate presentations, podcasts, and blogs from any document. Basically NotebookLM inside Acrobat.

• Spotify personal podcasts — your AI agents (Cowork, OpenClaw, Hermes) can now push AI-generated daily briefings straight into your Spotify library. Morning AI briefing on your car commute, fully automated.

• Apple AirPods with cameras are in late-stage development. Low-res visual context for Siri so it can "see" your surroundings.

• Nvidia + SPAN home AI data centers — new homes will ship with built-in AI compute modules (16 Blackwell GPUs each). Excess capacity may pay you back like solar panels did in 2010. Five years out, but worth tracking.

“🔥 Pro Tip: The voice agent stack OpenAI just shipped is the foundation for the single biggest agency play of 2026 — productized AI voice receptionists for local service businesses. Plumbers, HVAC, dentists, salons, law offices — every one of them is losing money on missed calls and after-hours leads. Build a GHL + Voice Agent package: install the agent, train it on the business's FAQ + booking calendar, configure handoff rules for hot leads, integrate it directly into GHL conversations. Charge $1,500 to $2,500 setup + $300 to $500/mo for ongoing optimization (the agent gets smarter with usage, the prompts need tuning, the business adds services). One agency owner running 10 of these is at $3,000-$5,000 MRR for roughly 6 hours of monthly work. The realtime voice models are API-only right now, but the moment they land in Cowork/Codex/GHL native (weeks, not months), you want this offer ALREADY positioned in your client base. Plant the flag now. Sell the upgrade later. “ —

Here's where this all lands. The pace isn't slowing — it's accelerating. OpenAI shipped 5 major things this week. Anthropic shipped 4. And we're not even at the year-end push.

The people winning right now aren't the smartest. They're the ones building services on top of these releases inside of 30 days, not waiting for the dust to settle. Voice agent builds. Office + Claude consulting. Cowork-based agency operations. There's a 90-day window on each before every competitor catches up.

Move.

😲 The New Claude Code Feature That Kills the "Wait, What Was I Doing?" Problem

Be honest. You've had six Claude Code terminals open at the same time. Each one running a different task. Each one demanding context. Each one slowly fading into the back of your brain until you suddenly realize 90 minutes later — "wait, did one of those finish? Did one fail? Is one still waiting for me to type 'yes' to a permission prompt I forgot about three hours ago?"

I've done it. You've done it. The cognitive load of running parallel Claude Code sessions is real, and until this week, the only "fix" was discipline (lol).

Anthropic just shipped Agents View — a unified dashboard for every Claude Code session you have running. Type claude agents in your terminal and the whole world calms down.

• Here's what it actually does:

  • One dashboard, all your sessions. Every active Claude Code session shows up in three columns — Needs Input / Working / Completed. You see at a glance which agents are waiting on you, which are still cooking, and which are done.

  • Click in or peek in. Click any session to drop into its full terminal view (back arrow returns you to the dashboard). Or just hover and hit spacebar to peek — see what's happening, type a quick reply, get back out. Like sticking your head into a room without committing to the meeting.

  • Bring existing sessions in. Already have Claude Code running in another terminal? Type /bg (background) inside it and it bumps straight into your Agents View. No restart, no losing context.

  • Sessions persist when you close the window. Reopen with claude agents and they're all still there, exactly where you left them. You're not going to accidentally nuke eight agents by closing the wrong tab.

  • Start new sessions from inside the dashboard. You can fire up new Claude Code tasks directly from the view. In theory, you never have to leave this screen again. Which is kind of the point.

  Real talk — this is one of those updates that doesn't sound that exciting until you use it once. Then you can't go back. If you've been running 3+ Claude Code sessions at a time, this isn't a productivity boost. It's a sanity boost.

  Fire up claude agents in your terminal this week. Watch your tab anxiety vanish in about 30 seconds.

💲Deals at a Glance

🔥 LIVE THIS WEEKEND — Build-A-Thon (FREE Registration) 2 days. Solo. Live. I'm building an entire GoHighLevel agency from zero — agency account, white-label setup, sub-accounts, snapshots, voice AI, chat AI, and the full services + sales playbook. No edits. No catches. Bonuses + special guests + live giveaways for everyone who shows up. Register Free →

🚀 Claude Code Workshop — 4 Days, 8 Hours, $67. Build Apps, AI Agents & Automations From Scratch. Register Now

Our GoHighLevel Affiliate Offer Best Affiliate Offer For GoHighlevel On The Planet. Get Started Now

🔥 Vibe Coding Incubator – Get hands-on with Vibe Coding, live weekly trainings, app templates, support, and Lovable AI How To's 👉 Join For Free

😲 Best Gohighlevel Course For Only $20! 1000+ Students. 4.7 in reviews. Over 18 hours of content: https://www.udemy.com/course/gohighlevel-masterclass-from-beginner-to-expert/?couponCode=GHLSPECIAL

Skool Community — Two Ways In Build your agency or side hustle with HighLevel, n8n, AI & Automation. Your path to quitting the 9-to-5 starts here.

• Join Free Community — free Skool access, free course, daily community

• VIP Upgrade — $47/mo — direct hot-seat support, advanced templates, inner circle access

PRDs by voice. Bug reports by voice. Ship faster.

Dictate acceptance criteria and reproductions inside Cursor or Warp. Wispr Flow auto-tags file names, preserves syntax, and gives you paste-ready text in seconds. 4x faster than typing.

Try Wispr Flow free

Remember, there’s always something new coming from the Automated Marketer YouTube Channel—new content drops every Tuesday & Thursday. Subscribe so you never miss the next big idea.

Keep watching, keep learning, and keep moving forward!!

Nuno