Not theory. Not tutorials. These are the actual tools and projects I work with every day — and how I'm using them to solve real problems for real companies.
Each tool has a specific job. Together, they form a system that runs an entire business — from planning to building to deployment.
My "second brain." Everything I know about my business, clients, processes, and goals lives here. I've built a structured vault that acts like a living knowledge base — organized by company, department, and project. AI can read and write to it, making it a true organizational memory.
N8N is the plumbing that connects everything. I use it to build automation pipelines — moving data between apps, triggering AI actions, processing form submissions, and syncing systems without writing a line of custom code for every connection.
GCP is where the work lives in production. I deploy AI applications on Cloud Run, store data in Cloud Storage, and use Gemini for reasoning. It's enterprise-grade infrastructure that scales up when needed and costs almost nothing when it doesn't.
Claude is my primary AI partner for complex reasoning, writing, and building. I use Claude's Cowork mode to manage daily business operations — drafting documents, analyzing data, making decisions, and working through problems the same way you'd work with a senior colleague.
Codex accelerates development by turning plain language into working code. ChatGPT handles everything from quick research to first drafts. Together with Claude, they form a full AI development team — one that doesn't need breaks and never blocks a PR.
None of these tools work in isolation. The real value is the system: Obsidian stores the knowledge, N8N moves the data, GCP runs the apps, and Claude/ChatGPT/Codex do the thinking. It's an AI operating system for a small business.
A typical workflow moves through multiple tools — each doing what it does best.
Most of this work is internal — built for my own business first. One is a deployed client project. All of it is real, running in production, and built the same way I'd build it for you.
The site you're on was built entirely by me using AI — no web designer, no agency, no template. Every page, every line of copy, every animation, the chatbot widget, the lead capture forms, the SEO metadata, the structured data markup, and the deployment scripts were all written with Claude and Codex, then shipped to a Google Cloud VM.
I described what I wanted, worked through it iteratively with AI tools, and deployed it myself. That's the same process I use to build client systems. This site isn't just a portfolio — it's a live demonstration of exactly what I do.
Running a business solo means you can't afford to lose context. I built a complete AI-powered knowledge management system in Obsidian that acts as both an organizational memory and an active decision support tool.
The vault contains structured knowledge across every business function — including a full virtual AI department with dedicated "employees" for Marketing, Sales, Operations, Legal, HR, Finance, and R&D. Each agent has a defined role, personality, and task list. When I need to think through a marketing strategy or draft a sales email, I work with the appropriate agent who already has all the business context.
This is what it looks like to run a lean operation with AI doing the heavy lifting — not replacing thinking, but making sure every decision has the right information behind it.
Every manual, repeatable task in my business is a candidate for automation. Using N8N as the workflow layer, I've connected the tools I use every day — Google services, AI APIs, communication tools, and data stores — into automated pipelines that run without me.
Examples include: automated client follow-up sequences, data sync between Google Sheets and project management, AI-generated content drafts triggered by new lead entries, and scheduled reporting that summarizes business activity. The goal is simple: spend time on work only a human can do.
A local roofing company needed a faster way to capture and qualify inbound leads — without adding headcount or changing how they operate. I designed, built, and deployed a conversational AI chatbot on Google Cloud Run that handles the entire intake process.
Homeowners can describe their needs, upload photos of damage, and submit a lead without ever calling the office. All lead data flows directly into a Google Sheet for the team to action. The chatbot is live and handling requests in production.