Rolling into Day 3 of the challenge 🙌 What I built: A custom "company brief" skill for Claude Code. I feed it a company name or a URL, and it does the research and spits out a genuinely sharp-looking HTML report, overview, products and services, market positioning, recent news and signals, all with real cited sources at the bottom. My sample run was on Republic Services, and it came out looking like something I'd actually hand to a client. It works for straight company research or competitor analysis. Then I got greedy and built a second skill that takes that HTML and renders it into a clean PDF. (Screenshots below 👇) One thing that clicked: Skills are basically reusable superpowers. Instead of re-explaining what I want every single time, I taught Claude Code how to do the job once, and now it just does it on command. The moment it really landed was chaining the two together; one skill builds the report, the next turns it into a polished PDF. That's when it stopped feeling like a one-off task and started feeling like I was building my own little toolkit that compounds. One thing I'd improve next time: A couple of things on the list. I want to tighten the source quality, so it leans on primary sources like investor pages and earnings releases over aggregators. And I'd love to add a competitor comparison mode that briefs two companies side by side in the same report, since that's really where the analysis gets useful for my business. Wild how fast this stuff comes together once you stop thinking in one-off prompts and start building tools you reuse. Anyone else in the group building custom skills? Drop what you're working on, I want ideas for skill number three. 🚀

I only spent around 30 minutes on this to get a basic understanding of how to create websites in Claude Code. I didn't add diverse animations or anything but I made sure the website looked good still. https://www.risedigitalco.com/ This is the frontend page for my AI Agency although there isn't a backend as I'm not quite finished with designing the website. The hack that helped the most from Nates website building video was definitely the front-end design skill. I tried making a website with Claude a while back and it looked absolutely horrendous, I wasn't using the front-end design skill then but now I did and there is a MASSIVE difference.

Good afternoon everyone. As I’m closing down on the 2nd chapter of the course, I wanted to know when do I start building in n8n? I been so eager to build but just been patient with learning through the videos.

New here and jumping into the 7-day challenge, so hi everyone 👋 What I built: An automated newsletter workflow. I hand it a topic, and it takes it the rest of the way: researches recent sources, drafts the copy in my brand voice, generates on-brand infographics, hosts the images, builds the email-safe HTML, and sends it to the inbox. Topic in, finished newsletter out. My first issue was on the rising cost of funerals and how final expense insurance helps. (Screenshots below 👇) One thing that clicked: The big lightbulb was how much you can hand off to the agent once the workflow is set up right. I give it a topic and it makes the calls the whole way down, research, stat selection, copy, layout, send. The trick is splitting the work into predictable stuff the tools handle and the judgment stuff the agent reasons through, so the whole thing runs start to finish without me hovering over it. Oh, and a fun little tell I learned to hunt down: those long dashes (the em dash) are a dead giveaway that something was AI-generated. So I had it strip them out so the writing actually reads human. 😅 One thing I'd improve next time: Three tweaks on my list. Bake the dash cleanup in so it's automatic every time. Add a step that double-checks the data points against their sources before they land in the draft. And build a quick compliance checklist into the flow so it catches missing disclosures or a wrong link on its own. Honestly, this was way more fun and way simpler to put together than I expected. Excited for Day 2. Anyone else building something automated? Would love to see what you all are cooking up. 🚀

Back for round two of the 7-day challenge 🙌 What I built: I hooked up the Firecrawl MCP server and put it to work scraping real data off the web. Started with a pull from Yelp to grab some leads, and that went smoothly, no complaints. Then I got ambitious and tried scraping product reviews, first off Google, then off Best Buy, and that's where things got spicy. One thing that clicked: Not every site scrapes the same. The Yelp pull was quick and clean, but the moment I pointed it at Google and Best Buy for reviews, it hit some real snags. I ended up having to restart the whole scraping session, and the second run did behave better, so sometimes the fix really is just to tear it down and start fresh. Good reminder that the tool working perfectly once doesn't mean the next target is going to be as friendly. One thing I'd improve next time: The big head-scratcher was speed. That second run took way longer than my earlier Yelp pull, and I honestly can't tell you why yet. So next time I want to actually dig into what's slowing it down, whether it's the site fighting back, the volume of reviews, or something in how I set up the request. Figuring out the "why" behind the slow run is my homework. Not as clean as Day 1, but I learned more from it wrestling with me than the easy win did. That's the point of the challenge, right? 😄 Anyone else run into scrapers that work great on one site and choke on the next? Curious how you all handle it. (The third page, for some reason, kept failing, so I just told Claude to run with the 2 pages we had.)