Hotel Heartbeat

Hotel Heartbeat is a weekly, AI-powered email that synthesizes a property’s key performance data and delivers three prioritized signals — each paired with the context behind it and a concrete, recommended action. It’s built for hotel leadership teams who don’t have time to go digging, but can’t afford to miss what matters.

The problem

Hotel leaders aren’t short on data — they’re short on attention. Between a dozen dashboards and a flood of metrics, the signal that actually deserves action in a given week is easy to miss. The goal wasn’t another dashboard. It was to do the triage for them: figure out the few things that matter most for this property, this week, and say so plainly.

What I built

Heartbeat runs an agentic process that reads a property’s latest performance data, decides which signals are most important, and writes each one up with its context and a recommended action. The discipline is in the restraint — the hard part of an AI insights product isn’t generating text, it’s deciding what not to send. Capping the output at three prioritized signals keeps every email focused and trustworthy; recommending an action (not just reporting a number) makes it useful the moment it lands.

A Python pipeline handles data processing and prepares the inputs; the agentic layer runs on Google Gemini 2.5 Flash; Node.js handles email templating and the reporting pipelines; and Braze delivers the campaign on a weekly cadence — all running on AWS. The result is property-specific intelligence that shows up on its own, every week, already prioritized.

Why it matters

Heartbeat turns a sea of metrics into a short, confident weekly nudge — the kind of focused, property-specific insight that supports faster, smarter decisions without asking a busy team to go looking for it.

Stack

• Google Gemini 2.5 Flash — agentic synthesis and signal prioritization
• Python — data processing pipeline
• Node.js — email templating and reporting pipelines
• Braze — email campaign delivery
• AWS — cloud platform