A real generation, replayed. The pipeline that produced this article ran against
OpenAI through Inngest, writing every step to a cost ledger in Postgres: which model,
how many tokens, how long, what it decided. This page carries that trace and plays it
back — the outputs, the timings and the costs are the recorded ones, and the
ledger is recomputed in your browser from tokens × the app's own pricing table.
The client's name is replaced with a placeholder; nothing else is edited.
The brief what a person asked for
Brand context the corpus and profile that shape every prompt
The pipeline outline → draft → images → QA → SEO
cost so far $0.000000
tokens in 0
tokens out 0
wall clock 0s
recomputing the ledger…