Clip'd Score any video free
Hype Check · Greg Isenberg

Worth your time?

How I use Claude Code + MCPs to run my marketing by Greg Isenberg

How I use Claude Code + MCPs to run my marketing Watch on YouTube
Hype score · lower is better
34/100
Useful but promotional

Teaches practical, hands-on methods for building marketing agents and ad pipelines, with useful tool names and workflows, while including light self-promotion.

Substance75
Proof45
Pitch pressure28
Claim inflation40
The 15-second version

Post it, send it, settle the argument.

Download clip
Receipts

Why this score

  • 2:16Speaker defines a marketing agent and lists concrete infrastructure (data pipeline, warehouse, cloud-hosted code), which boosts substance by giving clear, implementable components.
  • 4:46Caller states WordPress powers ~43% of the web and uses agency migration examples (Time, TechCrunch), a concrete market claim that raises perceived opportunity but lacks external sourcing, so it helps substance while only moderately helping proof.
  • 24:14Gives specific operational detail, 'two different adsets per day, five ads per adset' and the agent's winner/loser loop, this is practical workflow-level advice that raises substance and shows how it runs in production (improves proof moderately).
  • 18:53Host asks viewers to follow across channels and signals 'if you want more content like this,' which is a soft self-promo that raises pitch pressure without gating core content behind a paid product.
Title partialTitle promises 'Claude Code + MCPs' specifically; the video delivers marketing agent how-to and many tools (Perplexity, Airbyte, ClickHouse, HeyGen, SeedDance, etc.) but never discusses Claude Code or MCPs.
Skip the video

The short version

Kept by Clip'd

Cody Schneider explains how to build marketing agents: small pieces of cloud-hosted code that read unified business data, make decisions, publish creative, and iterate on results. He walks through a concrete opportunity (AI-first WordPress plugins), the ad-channel playbook (Facebook ads + Andromeda), the required infrastructure (data pipeline, warehouse, hosting, Facebook Marketing API), and practical tactics to avoid common failure modes like entropy.

Keep the lesson

What it teaches

  • Build a closed-loop agent when you can feed it reliable, unified data and measure outcomes. Use it to automate repetitive optimization (ad creative selection, pausing losers, scaling winners). Don't apply this before you have product-market fit or clean instrumentation; an agent without good signals just amplifies nois
  • Treat agents as domain-specific decision systems, not mythical general intelligence. Design them as small, testable decision trees that operate on concrete metrics (CTR, conversion, revenue) and host them in cloud code with access to your warehouse. Avoid turning this into a 'set and forget' campaign, plan human checkp
Jump straight there

Key moments

  • 2:16Cody gives a tight definition of a marketing agent: unified data, autonomous decisions, and a cloud-hosted thinking loop.
  • 4:12Pitch of the 'AI for WordPress' startup: bundle common plugins, add AI-first features, and sell tokens as subscription pricing.
  • 11:21Bold claim about channel strategy: Facebook has become the best B2B ads channel because the algorithm finds obscure niche problems.
  • 19:04Infrastructure prescription: use AirBite to build a pipeline and ClickHouse as a self-hosted warehouse to feed agents.
  • 24:44Agent operational example: publish many ad variations, wait 48-72 hours for signal, pause losers and promote winners automatically.
Lines worth keeping

Quotes

  • The hardest part about building a startup today is getting customers to the startup.0:00
  • Facebook has turned into the best B2B ads channel and that exists right now.11:21
  • an agent is just code under the hood, right? It's just a decision tree that's like, cool, here's a live data stream.23:47
  • You're going to know that within 48 hours.30:29
Clip'd

Skip the video. Keep the lesson.

Score your next video free