A quick look at what I'm building, the tools I use, and how it all fits together — from idea to live web app.
I'm not a professional developer — I'm someone who figured out how to build real, working web apps by combining the right tools. Here's the honest overview.
I use Claude (Anthropic's AI) to generate all the HTML, CSS, and JavaScript for my apps. I describe what I want, it writes the code. I review, refine, and ship it. No coding from scratch.
Instead of writing server-side code, I use Make.com to wire everything together — receiving form submissions, calling AI APIs, saving data, sending emails, and responding to users. All visual, no backend code.
Every app stores its data in Airtable. It's a spreadsheet-meets-database that Make.com can read and write to automatically. User submissions, AI outputs, usage logs — all in one place.
My website is plain HTML files stored in a GitHub repository. Every time I push a change, GitHub Actions automatically uploads the files to my web host. Changes go live in under a minute.
Here's the rough process for how a new app gets built and shipped.
I describe the app to Claude — what it should do, how it should look, what the user experience is. Claude generates the full HTML/CSS/JS in one shot. I iterate on it in conversation until it looks and works the way I want.
I build a Make.com scenario (a visual workflow) that receives data from the app, runs it through an AI model (like Claude or OpenAI), saves the result to Airtable, and sends the response back to the user — often with a follow-up email too.
The app's HTML sends a POST request to a Make.com webhook URL when the user submits a form. The webhook triggers the automation, which processes the request and returns a result. It all happens in seconds.
I drop the HTML file into my GitHub repository, push it, and GitHub Actions FTP-deploys it to Bluehost automatically. The app is live on my website within a minute of pushing.
Airtable gives me a live view of every submission and AI response. Google Analytics tracks usage. If something needs tweaking, I update the code or the Make.com scenario and redeploy.
Every tool here was chosen because it lets me move fast without needing a dev team or writing server-side code.
My primary AI for writing code. I describe what I want in plain English and Claude generates the full HTML/CSS/JavaScript. It also powers several of my apps as the AI engine running inside the Make.com scenarios.
The backbone of every app. Make.com is a no-code automation platform where I visually build workflows (called scenarios) that receive webhook calls from my apps, run AI, save data to Airtable, and send emails — without writing any server code.
Every app stores its data here. When a user submits a form, Make.com saves the input and AI output to an Airtable base. I can see all submissions in real time, filter them, export them, or pull them back into other automations.
All my site files live in a GitHub repository. A GitHub Actions workflow watches for pushes to the main branch and automatically FTP-deploys every file to Bluehost. Push code → it's live. No manual uploads, no FTP clients.
Standard shared hosting where the site lives. GitHub Actions deploys files here over FTP on every push. The .htaccess file handles HTTPS enforcement, clean URLs, browser caching, and legacy redirects.
Claude Code is a terminal-based AI coding assistant that I use inside VS Code. It can read my entire codebase, make edits across multiple files, and deploy changes — all through conversation. It's how I actually manage and update the site day-to-day.
Embedded on every page to track visits, page views, and user behavior. Helps me understand which apps get traction and where people drop off.
After processing a user's request, Make.com sends a follow-up email with the AI-generated result. No email service setup required — Make.com handles the send using built-in email modules.
For the more technically curious — here's exactly how a user request flows through the system.
Every app and page on the site is a standalone .html file. There's no React, no Vue, no build step, no node_modules. All CSS and JavaScript is written inline inside each file. When I need a new feature, I update one file.
This keeps things simple, fast, and easy to understand. Any browser can run them. Any server can host them. Claude can read and edit them directly.
Every AI-powered app follows the same 3-step automation chain:
Scenario 1 handles the user-facing response — it calls the AI, gets a result, and responds to the browser immediately. Then it fires an HTTP call to Scenario 2 in the background, which sends the user a formatted email with their result.
Every fetch call from the front-end to Make.com includes a secret token in the request payload. On the Make.com side, a Token Guard filter rejects any request that doesn't match. This prevents the webhook URLs from being abused by anyone who finds them.
Two apps — Voice Todo and Voice Todo Dashboard — are Progressive Web Apps (PWAs). That means you can add them to your phone's home screen and they work like native apps. They use service workers for offline caching and have custom app icons.
Service workers are registered at the root of the site so they have full scope. The strategy is cache-first for static assets (fast loads) and network-first for Make.com webhook calls (always fresh data).
The GitHub Actions workflow (.github/workflows/deploy.yml) runs on every push to main. It uses FTP-Deploy-Action with credentials stored as GitHub Secrets. The entire site is synced to Bluehost over FTP automatically — no staging, no manual steps. Push → live.
A growing collection of web apps, all built with this same stack.
A voice-powered to-do list that records your tasks by voice, transcribes them, and syncs everything to Airtable. Installable on your phone like a native app.
Open app →A companion dashboard for Voice Todo showing task analytics, completion rates, and a full task history pulled live from Airtable.
Open app →Enter a topic and get a full SEO-optimized blog post generated by AI. The result is emailed to you and saved to Airtable for review.
Open app →Parents enter a child's name and interests, and AI generates a personalized illustrated story. Designed for kids ages 4–10.
Open app →An AI reading companion for elementary students that generates comprehension questions and reading-level content tailored to the student.
Open app →AI-powered social media content generator. Input your topic and tone, get ready-to-post content for multiple platforms.
Open app →Generate platform-specific social posts tailored to your brand voice and audience with AI assistance.
Open app →A KPI and metrics dashboard that pulls live data from Airtable and visualizes it for quick business overviews.
Open app →I'm happy to walk through how any of these pieces work in more detail.
Get in touch