Claude Code parses wedding-concierge usage logs

Thin_Sky builds a wedding platform for guests travelling to Mauritius and later investigates how they actually use its AI concierge. In the primary explanation of the published infographic, the author identifies Claude Code’s role precisely: writing scripts to parse PostHog metrics and language-model conversation logs.

The agent-assisted parsing supplies material for a detailed report that Thin_Sky writes by hand. Another AI then turns that report into the image. Claude Code is therefore credited with log-processing work, rather than being labelled as the unidentified agent powering the whole concierge or the image generator.

The wider firsthand account gives the analysis a concrete purpose. Guests repeatedly try to jailbreak the concierge, and practical adoption problems appear around login, time-zone displays, arrival information and photo sharing. These are reported observations, not a controlled evaluation of wedding-planning accuracy or security.

The useful research pattern is moving from stored interaction evidence to a human interpretation before creating a visual summary. An infographic alone can hide the distinction between measured events and the story told about them. The source does not publish the complete parsing scripts, a reproducible metric specification or independently audited jailbreak results. This entry establishes the completed parsing-to-report workflow and its human checkpoint, without treating the graphic as proof that the concierge always behaved correctly.

Prompt
Suggested prompt (written by Agents In Use, not the author's)

Source: Reddit · Thin_Sky