RMS reasoning layer

The RMS already does the high-value rate optimization, and beating it on price is a small prize. The real gap is portfolio intelligence: which property is underperforming its submarket, where channel mix is drifting, which pace anomaly points to an operations problem rather than a pricing one.

How it works

The plan is a Claude reasoning layer sitting across the RMS, the rate-shop data, and the PMS. It reads the feeds the other revenue systems already produce and looks for the cross-property signals a single-property optimizer structurally cannot see. The roadmap is shadow mode first, then narrow overrides, then co-pilot on one property.

• Underperformer-versus-submarket detection.
• Channel-mix drift.
• Pace anomalies that flag an ops issue rather than just a pricing miss.

Planned, in Phase 0 scoping. Deliberately out of scope: always-on autonomous loops making live revenue decisions. The honest read is that current models are not there yet for open-world revenue feedback, so this earns trust in shadow mode before it touches anything.

This is where the individual agents become a brain. Once the data layer and the point systems exist, the marginal move is reasoning over them, which is exactly the layer a real revenue team would provide and the one an owner cannot staff.