Fast execution. Modern tools.

The working system behind AI products, no-code web apps, e-commerce systems, automation and MVP launches for businesses across the GCC — the tools I actually use, organised by what they do for me.

Idea → Plan → Design → Build → Test → Deploy → Operate → Improve

What I use, and what for.

AI & Coding

Where most of the engineering happens now.

Claude

My first stop for reasoning through a problem, drafting specs and writing — and the model behind most of my agent work.

Claude Code

The terminal agent I hand real tasks to: it reads the repo, makes the change, runs the tests and opens the pull request.

OpenAI Codex

AI coding for implementation, debugging and codebase analysis — the second agent I use to understand unfamiliar code, ship features and automate repetitive engineering work.

ChatGPT

Research, quick analysis and first drafts when I'm thinking out loud rather than building.

Cursor

The editor I open when I want to stay close to the code: inline edits, multi-file changes and reviewing what the agents produced.

Design & Product

From an idea to something people can click.

Lovable

My fastest path from idea to a working product: real UI, backend, auth and bilingual support in hours, then iterated with the client watching.

Webflow

Marketing sites and content-heavy pages, where a visual build with production-grade output beats custom code.

Shopify

The storefront-to-checkout platform I build e-commerce on, with themes, apps and automation around it.

Development & Backend

The code and data underneath every product.

GitHub

Source control for every repository, pull-request reviews, issues and the workflows that gate what reaches production.

Supabase

Postgres, auth, storage and edge functions — the backend under most of the AI apps I ship, including this site.

Infrastructure, Deployment & Automation

How things get to production and keep running.

Vercel

I use Vercel to ship production web products quickly, manage preview deployments, and keep the path from GitHub to production simple.

Cloudflare

DNS, CDN, security and edge infrastructure for my domains — the layer in front of everything I run.

Resend

Transactional and product email: form notifications, product emails and anything a system sends on my behalf, with delivery I can inspect.

n8n

Self-hostable workflow automation connecting products, AI models and internal operations, so the repetitive work runs without me.

Communication & Project Management

Simple, focused tools — on purpose.

HEY

Email built around screening and focus. It keeps my inbox small and intentional, which is how I want to communicate.

Basecamp

Project management, communication, tasks and client collaboration in one calm place — one tool for the whole project instead of five.

Notion

Docs, specs and the knowledge base behind projects — where requirements become written scopes.

How I build + work + deliver

The stack only matters as a system. This is how it fits together across the life of a product — the same loop whether it's a weekend MVP or a production platform.

Think & plan

Everything starts as a conversation: what the business needs, who it's for, what exists today. I turn that into a written scope in Notion — requirements, research, the architecture and the first set of tasks — and use Claude to pressure-test the thinking before anything is built. The output is an implementation plan a developer, an agent or I can execute without guessing.

Design & prototype

I don't spend weeks on mockups. The idea goes into Lovable as a working prototype with real screens, data and bilingual copy, so the client reacts to something they can click. Webflow covers marketing pages; Shopify covers stores. Design decisions get made on working software.

Build with AI

AI isn't a separate tool in my workflow; it's in the loop. Codex and Claude Code read the codebase before touching it, implement features, debug, refactor, write tests and review each other's pull requests. I use them to research technical approaches, map unfamiliar code and automate the repetitive engineering work — migrations, fixtures, boilerplate — so my time goes to decisions.

Build the product

The product itself is React and TypeScript on the front, Supabase for Postgres, auth, storage and edge functions, and the APIs and integrations around it — payments, AI models, email, third-party data. Cursor is where I work when I want to stay close to the code. Everything lives in GitHub from the first commit.

Ship & operate

GitHub to Vercel to Cloudflare is the whole production path. Every pull request gets a preview deployment, merging to main deploys to production, and Cloudflare handles DNS, CDN, security and the edge in front of it. Domains, performance and reliability are set up once and then monitored — not revisited every launch.

Automate

Once a product is live, the repetitive work around it gets automated: n8n connects products, AI models and internal systems; Supabase functions and cron handle scheduled jobs; Resend sends the email a system should send. The goal is fewer manual steps between an event and its outcome.

Communicate & manage

I run communication through HEY and projects through Basecamp, and I keep it that way on purpose. One inbox with screening, one place for tasks, messages, files and client collaboration — instead of accumulating SaaS tools that each need their own attention.

Deliver & improve

Shipping is the start of the useful part. I watch analytics and error reports, collect feedback from real use, fix performance before it becomes a complaint, and keep iterating — smaller releases, more often. Live products get better every month, and the automation around them keeps the cost of that improvement low.