- Automated Marketer
- Posts
- π²You Can Speak To Your CRM and It Will Do Things, Seriously...
π²You Can Speak To Your CRM and It Will Do Things, Seriously...
Claude Code + GHL MCP setup SOP inside. Claude Code Workshop Cohort 2 is open β grab it before it sells out.

200+ AI Side Hustles to Start Right Now
AI isn't just changing businessβit's creating entirely new income opportunities. The Hustle's guide features 200+ ways to make money with AI, from beginner-friendly gigs to advanced ventures. Each comes with realistic income projections and resource requirements. Join 1.5M professionals getting daily insights on emerging tech and business opportunities.
The Workshop That Sold Out Is Back β And This Time We Built a Marketplace
500 people showed up to the first Claude Code workshop. I wasn't sure what to expect β honestly, I half-expected a ghost town and a bruised ego. Instead, 50 of them became founders. Built real things. And about 80% of the group pushed through and came out the other side with a completely different understanding of what's possible.
The other 20%? Yeah, it got hard. Terminal errors, Windows PATH configurations breaking silently, moments where nothing made sense. Most of them stuck with it anyway. That's the part I'm most proud of.
Here's the thing about learning Claude Code β it's not like watching a tutorial and nodding along. It pushes back. It breaks. It makes you feel like you have no idea what you're doing, and then suddenly something works and you realize you just built a functioning application with a live URL that would've cost you $500 in developer time.
We're doing it again. April 27β30, 4:30β6:30 PM EST. Four days, eight hours total, two apps built and deployed by the time we're done.
Here's what this cohort covers:
Day 1 β Getting your environment set up correctly. VS Code, Claude Code, Node.js, Git, and the CLAUDE.md playbook that gives Claude persistent memory of how you work. This is the setup 90% of YouTube tutorials skip β and why most beginners spend their first week troubleshooting instead of building.
Day 2 β Power tools. You'll install 5 MCP servers (GitHub, Context7, Sequential Thinking, Playwright, Tavily), build your first custom skills, and set up a 5-agent AI dev team running in parallel. Reviewer, Debugger, Frontend Dev, Documenter, Tester β all for $0.67/day.
Day 3 β Deployment. Supabase for your database, GitHub for version control, Vercel for hosting. You'll watch a complete app get built and deployed end-to-end, then do it yourself.
Day 4 β Ship it. A guest security specialist (former Amazon senior engineer, rated #1 AWS security expert) walks you through the vulnerabilities AI-generated code creates by default β and exactly how to fix them. Then we build two apps live: a Lead Magnet Quiz Funnel that captures and segments leads automatically, and an ROI Calculator that shows prospects the cost of not working with you. Send that link before a sales call and let the math do the selling.
We're also launching a brand new marketplace in the next week or so β so what gets built in this cohort has a real home to go to.
This is $67. Eight hours. Two instructors who run their businesses on the exact system they're teaching. And it comes with 10 detailed SOPs, CLAUDE.md templates, an MCP safety guide, a full agent and plugin directory, and lifetime community access.
It sold out last time. It'll sell out again.
If you want the VIP upgrade β recordings from this cohort, every future cohort, and a stack of extras β that's available too.
Grab your spot before April 24th: go.vcinc.ai/claudecode


π² Ask Your CRM. It Does It. No, Really..
You know that moment where you're staring at a 47-tab Chrome window, manually entering contacts, booking appointments, creating invoices, and you think β there has to be a better way?
There is. And HighLevel just shipped it.
It's called Ask AI, and the short version is this: you can now talk to your CRM and it will actually do things. Not just answer questions. Not just look stuff up. Actually execute tasks inside your account β while you sit back and watch.
Here's what it can handle right now:
Create, update, and search contacts β including bulk imports. Tested it with a 50-contact list. 100 operations. Done in seconds.
Book appointments β it finds your calendars, asks which one you want, and schedules. Even catches errors on its own (like a calendar with no assigned team member) and fixes them.
Create invoices and products β tell it the product name and price, and it builds it, creates the invoice, and sends it. One instruction.
Add contacts to pipelines with tags β just say "put them in lead management under new lead" and it happens.
Write and publish blog posts β it searches for relevant info, drafts a full post, generates a cover image in your brand colors, and walks you through publishing step by step.
Build and manage AI agents, workflows, voice AI β this is still early, but the capability is already wired in.
Fetch reporting data, manage media storage, create brand boards β it's not just doing CRM tasks. It's starting to run the whole back office.
The thing that sets this apart from every other "AI in your CRM" play is that this isn't a chatbot with a help desk vibe. It's an operator. You give it a multi-step task β create a contact, tag them, book an appointment, invoice them, add them to a pipeline β and it lines up all five steps, shows you a confirmation screen, and executes in sequence. If something breaks, it tells you why and offers a fix.
One live test: a single voice command generated a contact, a tagged entry, a booked strategy call, a $15 product, and a sent invoice. In one shot. The only hiccup was a calendar missing an assigned team member β and the AI caught that itself, flagged it, and resolved it without being asked.
That's not software. That's a junior ops person who works 24/7 and never needs a Slack DM.
It's currently in beta β here's how to turn it on:
Go to your Agency account
Left-hand menu β Settings β Labs
Find Ask AI and toggle it on
Hit Activate Feature and enable it for all sub-accounts
If you're on someone else's white label, ask them to flip the switch. Seriously. This is not a feature you want to wait on.
π₯ Pro Tip: Most people are going to use Ask AI like a search bar β one question at a time. Don't. The real unlock is chaining tasks in a single prompt. "Create this contact, tag them, book a strategy call, and send them an invoice for $X." Stack the instructions and let it run. You can also open multiple conversations simultaneously, so you can have it working on a blog post in one thread while importing contacts in another. That's where the time savings actually compound.
The way agency owners spend half their day on manual CRM admin β that's about to become a thing of the past. Ask AI is early, but the infrastructure is already there. The question isn't whether this will change how you run your agency. It's whether you'll be using it in week one or catching up six months from now.


π€ Claude Code Is Now Talking Directly to Your GoHighLevel Account
If you've been building in Claude Code and managing clients in HighLevel, you've probably thought β these two things should talk to each other. Turns out, they can. And setting it up takes about 30 seconds once you know what you're doing.
The setup uses something called an MCP β Model Context Protocol. It's basically a secure bridge that lets Claude Code talk directly to your HighLevel account. No Zapier. No third-party middleware. No API spaghetti. Just Claude, your GHL account, and a config block.
Here's what Claude Code can do inside HighLevel once it's connected:
Pull contacts and read conversations β your AI can see who's in your CRM and what's been said
Move pipeline deals β tell it where a lead should go and it goes there
Post social media content and blogs β one sentence, done
Check calendars and book appointments β ask it what's available, it tells you, you confirm
Run discovery on your entire account β pipelines, tags, custom fields, all loaded into memory so it knows your setup
The setup is four steps. Create a private integration inside HighLevel settings, grab your location ID from your business profile, drop one config block into Claude Code, and run mcp in the terminal to confirm it's connected. When you see "ghl mcp connected" β you're live.
The official HighLevel MCP server has 39 endpoints available right now. That covers most of what you'd need for day-to-day operations. If you want to go deeper β closer to the 290+ endpoints that exist in HighLevel's full API β you can build and host your own MCP server. That's a more advanced move, but the infrastructure is there if you want it.
One thing worth flagging: don't give it everything. When you set up the private integration, you choose which permissions to grant. Be selective. Adding users, deleting records, editing billing β those are the ones that can cause real damage if something goes sideways. Grant what you need for the tasks you're running and nothing more. 65 permissions is plenty for most use cases.
The real play here is building an AI agent team around this. One agent β call her whatever you want β handles GHL operations. Another handles content. Another handles client comms. They can talk to each other and hand off tasks. You're not just automating individual actions anymore. You're building a team that runs without you in the loop.
π₯ Pro Tip: Before you run anything live, paste the setup command into a fresh Claude chat first and ask it to verify the code isn't malicious and is correct for your system. Takes 10 seconds and saves you from a bad copy-paste job breaking something in production. Also β store your PIT token and location ID in an .env file, not hardcoded anywhere. If you ever share a folder or push to GitHub by accident, you'll thank yourself later.
The ability to type one sentence and have an AI reach into your CRM, move a deal, book a call, and post a blog post β that's not a demo anymore. It's a Tuesday morning. Get the MCP set up and start building the team around it.
Download the SOP here: https://go.vcinc.ai/claude-ghl-vcinc


The AI Week That Broke the Internet (And Almost Everything Else)
π§ AI Insider News
The AI Week That Broke the Internet (And Almost Everything Else)
There's a lot to unpack from the past week in AI. Some of it is genuinely wild. Some of it is marketing dressed up as a crisis. And one story in particular has the entire AI community arguing about whether we're all about to get hacked β or whether someone just wants us to think that.
Let's get into it.
The Mythos Situation
Anthropic built a model called Claude Mythos. They're not releasing it. And the reason is⦠unsettling, depending on how you look at it.
Here's what they said in their own words: Mythos has reached a level of coding capability where it can surpass all but the most skilled humans at finding and exploiting software vulnerabilities. It's already found thousands of high-severity vulnerabilities β including a 27-year-old flaw in OpenBSD, a 16-year-old issue in FFmpeg, and multiple chained vulnerabilities in the Linux kernel. You know, the software running most of the world's servers.
So instead of dropping it on the public, Anthropic launched something called Project Glass Wing β giving access to a small group of cybersecurity specialists at major tech companies. The idea: use Mythos to find and patch vulnerabilities before less responsible actors get their hands on models this powerful.
Is there a marketing angle here? Probably some. This kind of announcement does wonders for pent-up demand. But here's the thing β the benchmarks are real. Mythos scored 24 percentage points higher than Opus 4.6 on SWEBench Pro. Nearly double on multimodal software engineering. This isn't GPT-2 hysteria. The model is legitimately that good, and Anthropic is legitimately nervous about it. Hard to argue with that call.
The Models You Actually Get to Use
While Mythos sits behind locked doors, a couple of notable releases dropped this week that you can use.
Meta's Muse Spark β the first model out of Meta's newly formed Super Intelligence Labs β made a splash. It's not the best at any single thing, but it jumped from dead last on the AI intelligence index to fourth place overall. Token efficiency is its real selling point β it should be significantly cheaper to run than GPT or Opus. Good general-purpose model. Watch the API pricing when it goes wide.
The one that actually deserves more attention: GLM 5.1 from ZAI. It's open-source, MIT licensed, downloadable from HuggingFace, and it's matching GPT-5.4 and Opus 4.6 on software engineering benchmarks. Let that sink in. An open-weight model you can download, fine-tune, and run locally is now trading blows with the best frontier models in the world on coding tasks. That's a massive deal for anyone building on AI infrastructure.
Other Updates Worth Knowing
Gemini Notebooks β basically Google's version of Claude Projects. Separate memory per notebook, syncs with Notebook LM. Rolling out to paid users now. Took them long enough.
Seed Dance 2.0 is now available in the US through Runway and CapCut. Best accessible video model right now, though the viral features (celeb cloning, IP generation) have been stripped. Still worth playing with.
HeyGen Avatar 5 β 15 seconds of footage to clone yourself. The lip sync isn't perfect yet, but this is getting uncomfortably close to seamless.
Anthropic's managed agents β a new feature in the Claude console that ties agents directly to tools like Notion, Slack, Asana, and Intercom. Early days, but the template library is already useful.
OpenAI's $100/month tier β a new middle option between Plus and Max, with 5x more Codex usage. If you're building heavily in Codex, worth evaluating.
Perplexity + Plaid β you can now connect your bank accounts to Perplexity for read-only financial tracking. Either very useful or very weird, depending on your trust level.
Cursor remote control β you can now prompt Cursor from your phone to run agents on your dev machine. Genuinely useful for anyone running long background tasks.
π₯ Pro Tip: The GLM 5.1 release is quietly one of the most important things that happened this week for agency owners building on AI. An open-weight model at frontier-level coding performance means you can fine-tune it on your own data β your SOPs, your workflows, your client patterns β and run it without per-token API costs. If you're already doing AI builds for clients, this is worth a serious look. Most people are going to sleep on it because it didn't come with a press release. That's your edge.
The real takeaway from this week isn't any single model or feature. It's pace. The acceleration isn't slowing down. Frontier models are getting better faster than anyone can patch the vulnerabilities they're finding, open-source is catching up to closed models in real time, and the tools available to agency owners right now β if you're actually using them β are nothing short of an unfair advantage. The question isn't whether AI is changing your industry. It's whether you're the one using it or the one it's being used against.


π²The Notes App That Doesn't Hold Your Brain Hostage, The Second Brain App!!
Every couple weeks I fall down a rabbit hole of something that has nothing to do with funnels or CRMs and somehow still makes me a better operator. This week's rabbit hole: note-taking. Specifically, why Obsidian might be the most underrated tool in a serious builder's stack.
Obsidian is not a notes app. It's a second brain. The core idea is simple β your notes live as plain markdown files on your own computer. No subscription lock-in, no proprietary format, no "sorry we're shutting down" email. You own the files. You can open them in VS Code, iA Writer, literally anything. In a world where every tool wants a recurring payment and your data hostage, that's quietly radical.
The bi-directional linking is where it gets interesting. You can link notes to each other and Obsidian builds a visual graph of how your ideas connect. Sounds gimmicky until you're three months in and you realize your note on client onboarding is connected to six other things you wrote without even meaning to. That's not a database β that's how your brain actually works.
It integrates with Claude. You can set it up to ask questions of your own notes, run deep research, and auto-populate note properties using AI. The person demoing this keeps their original thinking and AI-generated content in separate sections β which honestly is a philosophy worth stealing for how you use AI in general.
The learning curve is real but front-loaded. The biggest mistake people make is importing everything from their old notes app on day one. Don't. Start fresh, link your own thoughts, and let the structure emerge naturally. Chasing plugins on day one is also how you spend four hours customizing instead of thinking.
Obsidian isn't for everyone. If you want something that just works out of the box, this isn't it. But if you're the kind of person who's constantly capturing ideas across five different apps and losing half of them β this is the one tool that might actually fix that.
Give it a download. It's free. Worst case you've wasted 20 minutes. Best case, you finally have a place where your thinking compounds instead of disappears.

π²Deals at a Glance
π Claude Code Workshop β 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
π₯ Vibe Coding Incubator β
Get hands-on with Vibe Coding, live weekly trainings, app templates, support, and Lovable Ai How Toβs
π Join For Free
π²Best Gohighlevel Course For Only $20!
1000+ Student 4.7 in reviews and over 18 hours of content: https://www.udemy.com/course/gohighlevel-masterclass-from-beginner-to-expert/?couponCode=GHLSPECIAL
Join Our Free Skool Community and Course
Build your agency or side hustle with HighLevel, n8n, AI & Automation. Your path to quitting the 9-5 starts here.
Join Now
Marketing Tech Professionals Ready For Hire (Done For You Work)
MarTech professionals trained by me, ready to join your team for any marketing technology needs.
Explore More
VIP Office Hours
For those who just want simple questions answered by real knowledgeable human beings. Join our group office hours 4 days a week.
Access VIP Office Hours

Remember, thereβs always something new coming from the Automated Marketer YouTube Channelβnew content drops every Tuesday & Thursday. Subscribe so you never miss the next big idea.
Keep watching, keep learning, and keep moving forward!!
Nuno

