Day 2: The One Where I Finally Started the Blog
Today was three repos in one evening. A client analytics platform from scratch, a personal site I've been dodging for years, and a company website that needed its copy cleaned up.
The numbers
- 21 commits across 3 repos
- 1 personal site launched
- 1 client analytics dashboard + scoring engine built from zero
- 1 company website refined
- 2 blog posts published
Client analytics platform
New client. E-commerce brand spending on TikTok with no visibility into what's working. Built them a reporting dashboard with campaign breakdowns, weekly progress tracking, and a prototype front-end to validate the data before wiring up the real integrations.
But the real thing is the Breakout Probability Engine underneath. It scores every SKU by velocity, classifies products into tiers, and flags what's about to break out vs what's bleeding money. That's not a dashboard. That's a decision tool.
SKU scoring and classification is the kind of artifact that sells the next engagement. Real business logic, not just API wrappers.
Empty repo to branded, deployable product in one evening. The harness pattern from Day 1 is already repeating. Repo structure, operator workspaces, API scaffolding, dashboard skeleton. I'm building the same stack every time now.
cheeann.com
Three years. I've been meaning to build a personal site for three years. You know the loop. Pick a domain. Open Figma. Stare at a blank artboard. Wonder if you have enough to say. Close the laptop. Repeat next quarter.
Today I opened Amp Code and just asked it to build me one. Next.js, blog engine, animations, analytics. Two posts up. Deployed. The whole thing took less time than my average "planning how to start" session used to take.
The funny thing is the dashboard was objectively harder. Multiple APIs, scoring algorithms, real business logic. Shipped it with zero friction. Scoped it, built it, done.
The personal site? Technically trivial. Emotionally? Three years of "what if nobody reads this" piled on top of each other.
The agent didn't fix that. What fixed it was making the cost of starting so low that my excuses ran out of room. You can't procrastinate on something that's already done.
Groove
Smaller scope but important. Cleaned the AI tells out of the company site. Removed em dashes everywhere. Rewrote the support messaging to focus on team independence instead of follow-up hand-holding. Hid a blog post that wasn't ready. Small commits, but the site reads like a human wrote it now.
What's emerging
A pattern I didn't expect: build the full thing in one session, refine in the next. Day 1 was the company site blitz. Day 2 I went back and cleaned it up while simultaneously building two new things.
Perfectionism can't stall what's already shipped. You can only improve it.
Running total
- 50 commits, 5 repos, 2 days
- 1 scoring engine, 1 analytics dashboard, 1 blog engine, 2 websites live
- 0 blockers