Just having fun with the title. there's nothing breaking about this haha. But check this quote out. “I never use my go-to market tools. I just have my agent do it for me”… Just heard this in an interview with a SpaceX Engineer this morning. Validating to hear because… THIS is the world I’m convinced we’re headed to with landing pages and funnels. Not just agents helping with the work. There's another level to this. It's autonomous agents doing the work while we do something else. It looks like this: - You think through a strategy like: “launching a legal campaign to generate mass tort leads on meta. I need a conversion optimized funnel that asks (insert questions). Here are 3 companies that have spent millions of dollars on paid ads. Lead quality matters and I need compliance for TCPA locked down. This means integrating with TrustedForm, etc. Review the example funnels and refine the strategy then build me the form, page, ads, etc - then you literally step away. Take a call, go for a walk … Agent gets to work. - agent pings you for review. Feedback. Edits. Launch Then… - agent starts collecting data, learning, and that’s when the optimization loops kicks off. This moment is here. And I know this because this is my new workflow now. But here’s the thing. Only like 5% of people are probably doing it. Maybe less. Most people still don’t know what an MCP sever is. Even I sometimes call it an MPC sever. That’s how early we are. But soon I think the vast majority of people will be working like this. So… I’ll be sharing more insights and content on how to build and optimize lead generation funnels , pages, etc with agents as we continue to head in this direction. But I'll just leave this here. Grokbot + LeadCapture via MCP is game changing. Because the grokbot agent, or whatever you call it and work autonomously and that opens up a ton of use cases that pretty much change everything. More to come...

This is crazy. I connected LeadCapture to Grokbot via MCP and built a trucking accident funnel right from my phone in like 10 mins. 🤯 Now here’s here it gets insane. Grokbot can do things on its own. So after going live you could have it run on a schedule to pull data and so on. I’m still trying to wrap my head around the use cases here

Our MCP server is entering BETA, so I’m starting this thread to document different use cases, prompts, and things I learn along the way. I posted the initial MCP update here: https://www.skool.com/leadcapture/mcp-just-hit-staging-soon-agents-will-be-able-to-interact-with-your-leadcapture-account?p=324d6a03 It’s pretty mind-blowing to watch an AI agent interact directly with LeadCapture, review account data, analyze funnels, and make changes. The analysis side in particular is opening up use cases I hadn’t even considered. An agent can slice and dice your account data in ways that would normally take a lot of manual digging. I’m going to use this thread as a running prompt dump for things that work well. Some MCP use cases I’m testing: - Clone a funnel and make specific edits - Set up a routing rule - Build a lead quiz from scratch with X questions - Change styling across all funnel steps - Review all funnels and tell me which had the lowest conversion rate over the last 5 days - Audit my entire LeadCapture account - Import Meta or Google Ads conversion data and compare it against LeadCapture performance - Identify funnel steps with unusually high drop-off - Compare conversion rates across funnels, traffic sources, or date ranges - Find opportunities to improve conversion rates based on account data ⚠️ Important BETA note Getting an MCP server right is surprisingly difficult. Right now, MCP is strongest for analysis and account intelligence. We’re still expanding the tools it can access for editing, building, and making changes inside LeadCapture. That’s currently the biggest gap, so some prompts that involve actually changing your account may not work yet. That’s also the purpose of this Beta. We want to see what people actually try to make agents do so we can keep expanding the MCP tools around real-world use cases.

AI voice agents for lead follow-up and improving response time are catching on fast and I think there’s a HUGE opportunity here. But after building one myself (more on that later), I’m more convinced than ever that qualification and verification at the point of capture are going to become even more important. A few risks/edge cases I ran into while building my AI agent (and worth keeping in mind if you're doing this... or planning to) 1) Calling, booking appointments, or handing off leads that were never qualified Yes, the AI agent can qualify someone during the call. But now you’re wasting call minutes, tokens, compute, etc. on leads that could have been disqualified before the call ever happened. Better solution: qualify and route bad leads out at the point of capture so the agent only spends time on leads worth contacting. 2) Calling bad or unverified phone numbers This is probably the biggest issue I’m seeing. Someone submits a form with a bad number → the AI agent immediately starts calling it → now you’re potentially burning numbers and creating all kinds of downstream problems. If you’re automating outbound follow-up, verifying that a phone number is real and contactable before it reaches the agent becomes incredibly important. 3) TCPA/compliance gets more complicated In the US, there are rules around when and how leads can be contacted. If an AI agent is automatically making follow-up attempts, you need hard rules around things like: Lead location/time zone When calls are allowed Frequency of follow-up attempts Consent/opt-in data TrustedForm certificates and other compliance data Automation makes follow-up easier, but it also makes it easier to automate mistakes at scale. The experience of using AI agents to handle inbound calls and follow-up is honestly mind-blowing, and I’m really excited to unleash some of what we’ve been building when it's ready. But there are so many edge cases, pitfalls, and downstream consequences to think through.

Having way too much fun with AI video. I'm launching some ads this week and one thing I'm testing here is pain point amplification in a scenario is intentionally absurd, yet relatable. The clown one is very risky because it might insult some people. But gotta test it. The prompt I used to get videos like this: Clown prompt: Hyper-realistic, cinematic 4K video. Natural lighting. Shallow depth of field. Clean audio. Professional commercial tone with absurd humor. Scene Description: A clown wearing full clown makeup (red nose, subtle wig, painted smile) is dressed in a sharp mortgage broker suit (navy blazer, white shirt, tie). He’s sitting at a modern mortgage office desk with a laptop, phone, and paperwork. The office looks normal and professional — city skyline through the window — which contrasts with the fact that he’s clearly a clown. The clown looks exhausted, serious, and frustrated, not goofy. Camera starts as a medium close-up and slowly pushes in as he speaks. No on-screen text. No graphics. No captions. Just the character talking directly to camera. Performance Direction: The clown speaks calmly and honestly, like a real mortgage broker venting. Dry delivery. Slight frustration. Self-aware humor. Dialogue (≈12 seconds, spoken by the clown): Critical Instruction (Important): The clown speaks directly to the camera on screen. His mouth visibly moves and is lip-synced to the dialogue. No voiceover. No narration. No off-screen audio. The audio comes only from the character speaking in the scene. When I buy mortgage leads from the big vendors… I feel like a clown. They’re shared. They’re dead. Everyone’s already called them. I got tired of feeling that way. That’s why I use TapMyMortgage — exclusive, equity-enriched mortgage leads that actually make sense.” Ending Shot: The clown pauses, exhales slightly, and looks straight into camera. Natural room tone. Hard cut to black. Overall Mood: Absurd but honest. Funny because it’s true. Clearly aimed at mortgage professionals.