Five systems, one pattern.

01

Command center

Operational Intelligence Layer

Turning an agency that ran on chat threads into a system that runs itself

A hospitality marketing agency serving hundreds of restaurant clients ran almost entirely out of WhatsApp messages and disconnected software. Work was invisible. I built an operational intelligence layer that made the invisible work visible, then kept extending it until it became the command center the whole agency runs on.

02

Forms engine

Capture Once, File Everywhere

A forms automation engine that turns the same facts, entered once, into every document they need to land on

Every business that runs on forms pays the same hidden tax. The same facts get typed by hand into document after document. A skilled person becomes a copy machine, errors creep in, and everything slows down. I built a system that removes that tax: you capture the facts once, and it files them everywhere they need to go.

03

Web pipeline

One Site, Shipped

A tool that reads a restaurant's scattered web presence, rebuilds it on a fast foundation, and ships it all the way to production

Most tools that claim to rebuild a website stop at the part that photographs well. I built one that does the whole thing, from reading the old site to shipping the new one to production. Its hardest test was a nineteen-location group scattered across separate sites, consolidated into one correctly structured site that search engines could finally read as a single brand.

04

Creative ops

The Flyer Machine

How a marketing team producing fifteen flyers a week rebuilt its production line so on-brand flyers come out in seconds

A restaurant-marketing operation runs on flyers, around fifteen a week across dozens of brands, nearly all of it funneled through one overloaded designer. The problem was never design talent. It was everything around the design. I built a system that renders an on-brand flyer in about thirty seconds as a draft the designer approves.

05

Adoption lesson

Built Well, Never Adopted

Why a genuinely good design lost to a group chat, and what that taught me about the difference between building something right and getting it used

The best feature I ever shipped is one almost nobody used: a Meeting Room with live bilingual transcription and one-click tracked action items, genuinely good, that lost to a group chat the team already lived in. This is the case study about why a better tool does not win, and the gap between building something right and getting it used.