The '6-Month Project' Is Dead: How I Executed a Brand Revamp in 4 Weeks
TL;DR: I rebuilt my entire brand — visual identity, site architecture, content strategy, and deployment infrastructure — in four weeks using Agentic Workflows and sheer stubbornness. This is the real story: what broke, what surprised me, and what I’ll never go back from.
What the Old Version Looked Like
Let me be honest about where I started.
The old site was a Wix template I’d chosen in about forty minutes three years ago. The colors were “professional purple” — I don’t even know what I was thinking. The copy was written in the voice of someone who was trying very hard to sound like a PMM rather than actually being one. Every section had a stock photo of a person shaking hands or staring at a laptop in a way that communicated absolutely nothing about who I was or what I could do.
It was a placeholder I’d kept updating in small patches — a new post here, a retweaked headline there — instead of fixing the actual problem, which was that the whole thing needed to go.
I knew this for at least a year before I did anything about it. I told myself it was a six-month project. I needed to hire a designer. I needed to nail down the brand strategy first. I needed to find time. I needed a lot of things I kept not having.
What I actually needed was to stop waiting for perfect conditions and start building.
The Decision Point
The thing that pushed me over the edge was embarrassing, honestly.
I was in a conversation with a founder I respect — someone who’d built a genuinely beautiful, fast, well-positioned site — and he asked me for my URL. I gave it to him. There was a pause. “Nice,” he said. It was the pause that got me.
I opened my laptop that night and just started. No plan. No brand strategy document. No designer on standby. Just a terminal, an Astro starter template, and the decision to figure it out as I went.
That was the real shift. Not a project kickoff. Not a stakeholder alignment. Just: I’m doing this now, and I’ll work out the details in motion.
Week One: Architecture and Identity Crisis
The first week was all infrastructure and existential questioning.
I set up the Astro project, connected it to Vercel, pushed the first deploy. That part was fast. The identity crisis took longer.
The Vercel dashboard showed my new site next to my old one and they looked like they belonged to different people — which, in a way, they did. I had to make a decision about who the new site was for. Not audience demographics. I mean: what version of me was this representing?
I spent two evenings just writing. Not copy — notes. About what I actually believe about PMM. About what had changed in how I think about AI and GTM. About what I wanted someone to feel after spending ten minutes on the site. I didn’t use AI for this part. I needed it to be fully mine before I started building with anything else.
By the end of week one I had: a deployed (empty) site, a color system, a type scale, and a values document that I’d never publish but that informed every decision afterward.
The color system took an embarrassing amount of time. I ran through probably twenty palette combinations before landing on the dark navy base with Neo-Mint, Terracotta, electric violet, and acid lime. When I found the right combination it was physical — I felt it settle. That’s how you know.
Week Two: The Agentic Workflow Layer
Week two is where the tools took over the repetitive work and I could focus on judgment.
I built a Make.com scenario that took my values document and generated a first-pass content brief for each major site section — About, Work, Blog, Lab. Not final copy. Direction. The AI gave me angles I wouldn’t have considered, a few I immediately rejected, and two that became the actual live copy on the site.
I wrote all the real positioning copy myself. But the AI drafts gave me something to react to, which is faster than staring at a blank document. The friction of editing is different from the friction of starting. I always edit faster.
For the blog, I mapped out the full Agentic PMM series structure in one session — 21 posts, loosely planned by topic cluster, with alternating tactical and conceptual angles. This post is the last in that first batch. I used AI to pressure-test the structure: “Given this series arc, what topics are over-represented? What’s missing? What would a skeptical PMM want to see that isn’t here?” The feedback was useful. I restructured three posts based on it.
The GTM Infrastructure for the blog itself — content pipeline, markdown workflow, Git-based publishing, Vercel previews — was built in two evenings. Once it was running, shipping new content was a git push. That’s it.
Week Three: The Late Nights
I won’t romanticize this part. Week three was rough.
I was running on five hours of sleep because I’d underestimated how long the visual polish would take. Every time I thought a section was done, I’d see it on mobile and something would feel off. The spacing. The hierarchy. The way a Terracotta button read against the dark background at 375px.
There’s a specific kind of tired you get from doing detailed visual work past midnight that makes you question every decision you’ve ever made. I almost reverted the whole color system at 1am on a Wednesday. I’d been staring at it too long and suddenly it looked garish and wrong.
I saved the file, closed the laptop, and went to bed. The next morning it looked fine. More than fine. This is one of the most useful things I’ve learned about creative work: your 1am judgment is bad. Sleep is not procrastination.
I also hit a real technical snag in week three: a component I’d built for the Lab section was causing a hydration error in production that didn’t appear in development. I spent four hours on it before finding the issue (a conditional render that Astro’s partial hydration handled differently than I expected). This is the kind of problem that would have stopped me cold a year ago. It was annoying now, but solvable. That gap — between “this stops me” and “this costs me an afternoon” — is the whole story of building technical fluency over time.
Week Four: Launch and the Feeling After
I pushed the final version to production on a Tuesday afternoon, then sat there for about ten minutes not doing anything.
It wasn’t dramatic. There was no big announcement. I updated the URL in my email signature, posted one thing on LinkedIn, and went back to work. But something had shifted.
The feeling wasn’t pride exactly — or it wasn’t just pride. It was more like: I proved to myself that I don’t need to wait for the right moment or the right team or the right budget. I can build the thing. The full thing. From identity to infrastructure to content strategy to deployment.
That’s the proof point that no case study or credential can replicate. You did the thing. You know you can do the thing. And now the bar for “I’ll wait until it’s perfect” gets much lower, because you’ve seen what good-enough-and-shipping actually looks like in practice.
What This Proved — Beyond the Tools
There’s a version of this story where the lesson is “AI makes everything faster, use these tools.” That’s true but it’s not the point.
The real lesson is about permission. I spent a year not rebuilding my brand because I was waiting for conditions that were never going to arrive simultaneously. The right budget. The right designer. The right block of time. In reality, I had everything I needed. I just hadn’t decided to start.
Agentic Workflows and AI automation made the execution faster. But the decision to start — at 10pm on a Tuesday with nothing but a terminal and a strong opinion — that was entirely human. That’s the part no tool can do for you.
The six-month project isn’t dead because of AI. It’s dead because you decided it was.
Key Takeaway for PMM Leads
The technical tools matter less than the decision to start before conditions are perfect. Agentic Workflows compress execution time dramatically — but the leverage is only accessible to the person who’s already moving. Stop waiting for the project to be sanctioned. Build the thing.
FAQ
Q: What tools did you actually use for the brand revamp? Astro 6.0 for the framework, Vercel for deployment, GitHub for version control, Make.com for content pipeline automation, and a mix of Claude and ChatGPT for copy direction and structure. The visual design was done in CSS — no Figma handoff, just direct code.
Q: How do you manage a project like this without a dedicated PM or designer? By making decisions faster and holding them loosely. When you’re the only person who needs to approve something, you can move quickly. The tradeoff is that you have no external check on bad decisions — which is why stepping away and sleeping on it is important. You are your own design review.
Q: What would you do differently if you did it again? I’d write the values document first and spend more time on it. Everything else flowed from that. I wasted two days on visual decisions that resolved themselves once I was clearer on what I was actually trying to communicate.
Q: Is four weeks realistic for someone with less technical experience? With Astro, Vercel, and a starter template, an eight-week timeline is very achievable for someone new to the stack. The first week has a learning curve. After that, it’s mostly content and judgment — which is exactly what PMMs are good at. Don’t let the technical surface scare you off. The actual hard part is the clarity of thinking, not the tooling.