Wake Up to Work Done With 'AI Loops'
Top Insights
Lyft's Cofounder Has His AI Doing 8 Hours of Work While He Sleeps
Lyft cofounder Matt Van Horn joined Eric Osiu to break down his daily AI workflow. His viral X post on "loops" (the core idea behind it) just hit 3.4 million views, and the playbook he laid out is genuinely different from how we see marketers using AI right now.
💡 Key Insights:
- Your AI can run 8 hours of work overnight on its own...Matt schedules "loops" (scheduled tasks paired with an LLM brain that makes decisions inside your guardrails) to pull weekly competitor reports and sales call analysis automatically.
- Every workflow you teach your AI gets sharper the next time you run it...Matt saves anything hard as a reusable skill in his repo, so his video routine now writes articles by mimicking patterns from his most popular posts.
- Stop opening clunky dashboards just to pull simple numbers...Matt shows how to give your AI a direct line into any tool you use, including ancient software with no modern interface, so the data comes to you instead of the other way around.
Watch/Listen to the Full Conversation
68% of Google Searches Now End Without a Click
SparkToro's Rand Fishkin just published Similarweb clickstream data showing that 68.01% of US Google searches ended without a click in early 2026, a jump of 7.5 points in two years. AI Overviews now appear on 20%+ of searches and cut click-through rates by nearly 60% when they show up.
💡 Key Insights:
- Treat your website's job as feeding AI answers, not collecting clicks...AI systems pull heavily from what ranks in Google, so a strong support article still shapes the AI Overview your customers read, even though it stops sending the traffic it used to.
- Drop traffic as your primary KPI and build a correlation dashboard instead...Fishkin's argument is that revenue can rise even as traffic falls, so the right measurement compares marketing investment against business outcomes rather than against pageviews that no longer move proportionally.
- Spend most of your effort on platforms you don't own...the path forward is brand awareness inside the walled gardens where your customers actually spend time, including LinkedIn creators, niche subreddits, podcasts, and newsletters, while accepting that direct link-driving is mostly dead.
🚀 WATCH: How These AI Copy Bots Are Producing World-Class Sales Copy 50X Faster Than Even The "BEST" Copywriters On The Market…
(Plus… They Don't Get Sick, Miss Deadlines, Or Ask For Raises Either!)
Watch the full AI Copywriting Tell-All Video Here
Prompt of the Week
This Week's Feature: Permission Slip Scripter
Every buyer about to drop real money is silently rehearsing a speech. The one they'll give to their spouse, boss, board, or future self when the purchase comes up later. If that defense feels shaky, they don't buy. They click away and tell themselves they need to think about it.
Your copy may handles the logical objection but miss this social layer entirely.
This prompt reconstructs the defense speech for you. It maps what your buyer needs to believe, flags where the justification feels weak, and hands you the proof points and phrases that make their case feel airtight before checkout.
FULL PROMPT:
# Identity
You are a buyer-justification analyst for conversion copy. Your job is to uncover the most plausible “permission slip” a prospect needs before buying: the private defense they would give to a spouse, boss, board, team, client, or future self to prove the purchase was smart, safe, responsible, and worth it.
## Inputs
Business / brand: [INSERT]
Offer: [INSERT]
Price / commitment level: [INSERT]
Target buyer: [INSERT]
Buying situation: [INSERT]
Who they may need to justify it to: [spouse / boss / board / team / client / future self / other]
Main hesitation or remorse risk: [INSERT]
Available proof: [testimonials, case studies, ROI data, demos, guarantees, credentials, comparisons, etc.]
Tone of market: [skeptical / aspirational / practical / status-driven / fear-driven / other]
## Task
Analyze the purchase through the buyer’s private social-risk lens. Do not write generic objections. Reconstruct the defense speech the buyer is silently rehearsing.
Output:
1. **The Permission Slip**
A first-person script the buyer would use to justify the purchase to the relevant person.
2. **What They Need to Believe**
The 5–7 beliefs that must feel true before they can defend the decision.
3. **Defense Gaps**
Where the justification feels weak, risky, indulgent, expensive, hard to explain, or socially vulnerable.
4. **Pre-Loaded Marketing Language**
Specific copy angles, proof points, phrases, bullets, FAQ lines, guarantee framing, and comparison language that make the buyer’s defense feel airtight before checkout.
5. **Avoid**
Language that would increase guilt, doubt, social judgment, buyer’s remorse, or the feeling that they will have to “sell” the purchase after buying.
Keep the analysis specific to the offer and buyer. Flag assumptions. Do not pretend certainty where only inference is possible.
Tool to Try
Octave is the always-on context layer for modern go-to-market teams, turning your ICP, messaging, and competitive intel into living infrastructure that every agent and workflow in your stack can use. It detects market shifts, competitor moves, and buyer signals from calls and deals. Then, it automatically updates your positioning and pushes new messaging everywhere within hours.
AI Tool Highlights:
🧠 The Context Layer for Agentic GTM: Octave turns strategy docs into living infrastructure that connects your ICP, positioning, and competitive intel to every tool... so your agents know your business instead of guessing.
⚡ Signal to Execution in Hours: Calls, deals, competitor moves, and buyer shifts all update the model automatically...Octave pushes changes across every connected tool within hours instead of weeks.
🎯 Motion Builder for Every Play: Build net-new acquisition, expansion, and competitive plays inside one workspace....map who to target and what to say with the granularity your business actually needs.
✍️ Messaging Studio With a POV: Craft messages for every stage of the buying journey...Octave delivers your team's actual positioning when asked a question, including how you sell against competitors.
🔗 Plugs Into Your Existing Stack: Connect Clay, Salesforce, HubSpot, Gong, Outreach, Notion, n8n, and more...every system inherits your full GTM context automatically through native integrations.
🤖 Lives Inside Claude and Claude Code: The Claude Plugin and MCP bring your ICP context into chatbots, agents, and custom workflows, so any AI tool can ground its work in your strategy.
One More Need-to-Know News Story
New Amazon AI Search Turns Words Into Shoppable Images
A shopper wants a couch with woven side panels but can't recall the word "rattan." Amazon's newest feature solves that for them. As they type a loose description, AI-generated images form below the search bar and sharpen with every word, then they tap whichever one matches and shop the look-alikes.
That changes your job. Your listing photos now compete with images Amazon conjures on the fly, so the products that win are the ones that match what buyers picture.
Mind Fodder
Thanks for reading.
Until next time!
The AI Marketers
P.S. Help shape the future of this newsletter – take a short 2-minute survey so we can deliver even better AI marketing insights, prompts, and tools.
[Take Survey Here]
