Independent consulting · February 2026
Yorktowne Apparel: one AI dashboard for a business scattered across systems.
A custom-apparel business ran on a collection of disconnected systems — inventory in one place, pricing in another, sales orders in a third. The owner had no single view of his own business. I built him one, with an AI chat interface on top.
The problem
Every question that mattered — what’s in stock, what did we sell, how are prices trending — required logging into a different tool and stitching the answer together by hand. Several of the systems had no API at all, so the data couldn’t even be exported cleanly.
What I built
- A central database aggregating inventory, prices, and sales orders across all of the disconnected systems — the first unified picture of the business
- Custom Playwright-based scraping for the systems with no API, keeping the central data fresh without changing how the business operates day to day
- An MCP server exposing the unified business data to LLM agents as structured, queryable tools, with OAuth 2.1 securing agent authentication
- An interactive AI chat dashboard where the owner asks questions in plain English and gets answers grounded in his real numbers
- An AI-driven gatekeeper that routes each request to the appropriate LLM agent based on its complexity — quick lookups stay fast and cheap, deeper analysis goes to stronger models
The outcome
The owner has centralized insight into his business for the first time — one conversation replaces a tour through half a dozen logins, and he can see comprehensive sales rep performance for the first time. And because the data layer is unified behind an MCP server, new agents and new questions don’t require new plumbing.