john mark lowry
§ ai tools · 2026

trafficstation mapper

Broadcast ad ops: station extraction from media buy reports — with the AI reserved for the genuinely unknown.

team one internal tool, deployed in production on the agency's internal aws/eks platform.

the short version

Extracts station call letters from media buy reports and pairs them with broadcast creative allocations to build dub/ship orders — work that ran on manual spreadsheet surgery before. Two modes (station mapping; market intersection across files), styled XLSX out, and a market-alias table that speaks the dialect of buy reports.

Two design choices carry it: a growing deterministic parser handles known file formats with zero model calls, reserving the LLM for ad-hoc PDFs — and hash-based deduplication is a product feature, not an optimization: any file processed by anyone on the team is a cache hit for everyone, surfaced in the UI.

Full case study in progress — artifacts pending clearance or writing time. The insight below is already earned.

Deterministic-first: known formats parse without a model — instant, free, unit-testable. The AI handles the ad-hoc formats, and the roadmap is to need it less.

the insight
ran onReact 19 + Vite · Express + TypeScript · PostgreSQL (hash-dedup file cache) · LLM extraction (ad-hoc formats only) · production on AWS

← all work