A Week in The Life of a Claudepilled Marketing Founder

Over the past month I’ve been claudepilled hard and have become a power user of claude code in vscode . . . but not to work on software development.

I’m the marketing/business cofounder of a multi 7-figure, bootstrapped SaaS company. Others have used these new AI skills to move more into product/dev, but that isn’t me. I still am not touching product at all. 100% of my claude code work is marketing or running-the-business stuff like financial projections.

So I thought it would be fun to ask claude everything we accomplished in the past week. Here’s what claude provided:

15 YouTube videos produced and published

This was the big one. We went from zero video capability to a full repeatable pipeline in a week.

The pipeline we built:

• Blog post or workshop recording goes in
• AI writes a narration script optimized for YouTube retention and Google search
• ElevenLabs generates the voiceover
• Claude writes a Python renderer that creates animated slide videos synced to the audio (word-level timing via Whisper)
• Claude generates the title, SEO description, tags, thumbnail, and pinned comment
• A bulk upload script pushes everything to YouTube with thumbnails and playlists

What actually shipped:

• 14 short educational clips extracted from two workshop recordings, each turned into animated slide videos with Whisper-synced text
• 1 blog-to-video conversion (full 8-minute narration from a blog post, TTS audio, 20-slide animated video)
• All 15 uploaded to YouTube with SEO-optimized metadata and custom thumbnails

The slide video skill went through massive iteration — started with 4–8 sparse slides that sat static for 30–40 seconds, ended with 15–20 dense slides with something new appearing every 3–5 seconds. Every animation delay is derived from Whisper’s word-level timestamps, not guesswork.

Google Ads campaign rebuilt from scratch

Analyzed thousands of search terms to find what was actually converting vs. burning money. Built 15 new ad groups with targeted keywords, wrote all the ad copy, created 150+ negative keywords to stop waste, and executed the whole thing through Google Ads Scripts (automated, not manual clicking).

Also wrote an automated “Negative Keyword Catcher” script that runs on a schedule to find and block bad search terms before they waste money.

Pinterest strategy from zero

Audited the existing Pinterest account, identified 5 problems (no visual consistency, heavy repinning hurting reach, zero video pins, missing our best keyword clusters, descriptions that weren’t SEO-optimized).

Built a full strategy doc, then executed:

• Left/archived group boards that were diluting account quality
• Renamed boards to keyword-targeted names with SEO descriptions
• Created 5 new boards targeting high-value search terms
• Rewrote the profile bio for search
• Configured Tailwind scheduling (35 pins/week, automated)

Next phase: AI-generated pin templates so we can turn every blog post into 3–5 unique pin designs automatically.

Ad creative and strategy

• Designed 12 retargeting ad images (testimonial-based, 4 sizes each for different placements)
• Generated 9 ad creatives across 3 different visual formats
• Developed a new emotional angle for our lead magnet ads (imposter syndrome framing vs. the existing rational/data angle)
• Overhauled our creative testing methodology — changed from too many small-budget tests that never exit learning phase to fewer, properly-funded tests that give a clear read in 7–10 days
• Built lookalike audiences from CRM data

Cross-platform ad analytics infrastructure

• Created weekly deep-dive skills for each ad platform (Meta, Google, LinkedIn) so analysis is consistent and repeatable
• Built benchmark docs with platform-specific CPL ranges, conversion rates by time horizon, and spend ceilings
• Modeled maximum ad spend ceiling from P&L data and built a phased scale-up plan
• Fixed a bunch of spreadsheet formulas that were silently returning wrong numbers (wrong column references, data connector in append mode duplicating rows, missing ad set IDs in formulas)

LinkedIn tracking fix confirmed

We’d identified a bug where ~34% of paid LinkedIn leads were misattributing due to hardcoded UTMs in our link shortener overriding LinkedIn’s dynamic tracking. After fixing the links, verified that 100% of new leads are now tracking correctly.

Instagram carousel automation

Built an n8n workflow that scouts trending Instagram carousel formats, adapts them to our niche, generates slide copy and images, and posts a draft to Slack for review. Also manually built a second carousel in Canva via the API.

Infrastructure

• Built a Google Docs reader (so Claude can pull content from our Google Docs directly)
• Set up Google Ads API access for future automation
• Added a mandatory self-review system — Claude now runs through a checklist before declaring any code “done” (this fixed a pattern where it would write code, say it was finished, then I’d find bugs on review)
• Cleaned up and compressed the instruction files that load every session — cut context by 48% with zero information lost

How this actually works

[ Claude had the idea to add the section below, and it answers a question I get asked a lot so I’m including it . . take it away claude! ]

Claude Code runs in my terminal. It has access to my file system, APIs (Google Sheets, ActiveCampaign, Slack, YouTube, ElevenLabs, Canva), and can write and execute code. I describe what I want, it figures out how to do it.

The “skills” are reusable instruction files that capture what we’ve learned. When I say /slide-video, Claude loads a detailed skill doc that tells it exactly how to plan slides, sync timing to audio, render frames, and verify the output. These skills accumulate — each session's lessons get folded back in so the same mistakes don't repeat.

The key insight: most of this isn’t “AI doing my job.” It’s AI handling the production work (rendering videos, writing upload scripts, formatting spreadsheet formulas, generating image variations) so I can focus on strategy decisions (which angles to test, where to allocate budget, what content to produce).