Submitting 2 challenges in one day :) Day 3 is about skills. Mine is called client-voice-dna. It lives in my .claude/skills/ folder, it's 92 lines of markdown, and it sits in the middle of a chain that ships actual paying-customer content for two of my businesses. The honest version of this post is: the skill itself isn't the impressive part. The impressive part is the sequence of skills it's wedged between. Day 3 for me is really about what happens when you stop building individual skills and start composing them. Here is the chain (see attached dashboard): Voice notes and Reels → Wispr to transcribe → client-voice-dna (this skill) → Prism (Claude agent that drafts copy) → humanizer skill (strips AI tells) → ship. What client-voice-dna does It reads a transcript and produces a two-register voice DNA profile for a specific person. Public voice on one side: confident, in their vocabulary, ships as copy. Private voice on the other: hedges, self-deprecation, friend-register fillers, reference only. The output is a markdown file saved at active/<client-slug>/<client>_voice_dna.md. How I trigger it Whenever new client source material arrives. A new Sharpr client sends WhatsApp voice notes, that's a trigger. A CreatorsForge partner records a batch of Reels, that's a trigger. A discovery call transcript drops, that's a trigger. It always runs after the speech-to-text pass and always before any copy gets drafted. Two real uses, two different businesses Sharpr Automations is my AI automation agency. Clients hire me to wire automations into their business, and a lot of those automations produce content: captions, DMs, masterclass invites, email sequences. None of that ships unless it sounds like the client themselves wrote it. My first Sharpr client is a UK-based stretching coach. He records WhatsApp voice notes for me at the end of his day: what he's seeing in his sessions, what's working, what he wants to teach next. I run them through Wispr to transcribe, the skill builds the voice DNA file from the transcript, and from then on every caption, masterclass invite and DM that ships for him is drafted by Prism (the name I gave my Claude agent) against that file. The copy reads like he wrote it, because at the level of vocabulary and cadence and signature lines, he did.

I've spent over 400 hours inside Claude, and I'm breaking down exactly what separates someone stuck on level 1 from someone running five parallel sessions while they sleep, with the cheat codes to jump between each stage. Hope you enjoy!

Day 7 was a great learning experience as I successfully built my own personal executive assistant. One of its first capabilities is identifying leads within a specific domain and generating personalized first-outreach messages for sales and client engagement. Excited to continue improving and expanding its capabilities further.

Hey everyone! Here's my submission: https://fuellab.sharpr.cloud/newsletter/issue-001.html Quick intro: I'm a software engineer, Level 3 PT and Level 4 Advanced Sports Nutritionist, HYROX racer, half marathon runner, and getting ready for my first Ironman next year. You also saw my recent winning post on the community a few weeks ago: https://www.skool.com/ai-automation-society/community-wins-recap-apr-25-may-1?p=2b0f460e So when I started looking at what to build for Day 1 of Nate's 7-day challenge, the project that picked itself was tied to the thing I'm actually trying to launch on Skool — a fitness nutrition community for athletes and busy people who train hard. https://www.skool.com/test-community-3892 Why this was the right project to start with Soft launch is close. To get there I needed a landing page, an email list, a video sales letter on that page, the legal pages every site needs, and a way to actually publish weekly content without burning a full day every Wednesday. Nate's Day 1 brief said "build a newsletter automation." I figured I'd kill a flock of birds with one stone — newsletter pipeline, landing page, subscriber capture, VSL, and the legal footer, all in one connected system, all in one session. The stack I'd been wanting to try Claude Design for a while, so the landing page got built there first — pure React + Tailwind, no build step, the hand-drawn-coach aesthetic I'd been carrying around in my head for months. That gave me a real surface to attach the newsletter to. For the video sales letter, I used HeyGen — recorded one solid take, let it handle the avatar lipsync and the cleanup. The footage embedded straight into the landing page hero. Saved a half-day of trying to film myself in a quiet room. For the Privacy and Terms pages — every site needs them, nobody enjoys writing them — I just asked Claude. It produced two real, brand-appropriate pages that hold up. Five minutes from "we need legal pages" to "legal pages are live."

Day 2 of the 7 day challenge. Claude Code and MCPs are already my daily drivers, so neither of those was new tonight. The new piece was Firecrawl, and to explain why it landed the way it did, I want to walk through the side of my world that has been heavy on scraping since the end of last year. A business I run called CreatorsForge (https://creatorsforge.co). CreatorsForge is a Shadow Operator setup. I find Instagram creators who have built real, engaged audiences but who have no monetisation in place. No products, no offers, no funnel. I partner with them on a revenue share. They keep posting, and behind the scenes I build the entire business under their name. Their brand is the front. My operation is invisible. The AI involvement is different at each price tier. - For low-ticket, Claude generates the digital product itself; the creator advertises it to their followers, who buy on my platform, and the whole thing rides on a 14-day launch strategy my Claude agents put together. - For mid-ticket and high-ticket, the creator records everything themselves. They are the expert in their niche, I am not, and that's the whole point. The videos and the 1:1 coaching live on my platform; my job is the launch strategy and the infrastructure, not the content. None of this works without knowing two things cold. The creator's audience: where they hang out, what they actually want, what language they use. And the creator themselves: how they talk, how they phrase things, the cadence of their voice. Because every digital product Claude generates has to sound like the creator wrote it, not like a generic AI did. For both of those jobs, since early this year, I've leaned on Apify. I use one Apify actor to find the right kind of creator: around 10,000 Instagram followers, strong engagement, comments full of buying intent and no link in bio. Once we're working together, two more actors do the heavy lifting. One pulls every Reel and Post the creator has ever published and extracts the transcript captions, which feed a Claude agent that builds a voice DNA file: vocabulary, cadence, recurring phrases, the way they open and close. The other scrapes the comment threads at scale, and a different Claude agent clusters the pain points and surfaces the audience's actual language. Voice DNA on one side, audience pain language on the other. Both feed the product design and the launch copy, which is how a low-ticket digital product can plausibly read like the creator wrote it themselves.