LiteAPI CommandDEMO
DEMOIndependent demo built on LiteAPI's public sandbox and open-source MCP server. Not affiliated with Nuitée.
source

AI Recommendations · nightly agent output

An LLM agent reads the joined supply × pricing × bookings × MCP tables every night and ships a ranked next-actions list. The same agentic loop you already open-sourced — turned inward on your own data.
P1 actionsi
2
P2 actionsi
2
P3 actionsi
2
Avg confidencei
82%
P1Pricingconfidence 86%
Hilton Barcelona is over-priced by 5.5% vs market
LiteAPI rate €268 vs market min €248. Likely losing ~14 bookings/week to competitor 0.3 km away.
Action: Adjust take-rate or push rate negotiation to property
Impact: +€18k/quarter
P1Supplyconfidence 79%
Lisbon midscale supply gap (3* hotels under €150)
Search-to-shop ratio for 3* under €150 in Lisbon is 28% (vs 41% network median). 47 hotels match but unmapped.
Action: Run property-mapping batch on EAN feed for PT 3* segment
Impact: +€32k/quarter
P2Agenticconfidence 91%
MCP /hotels/rates p95 trending up 28% over 14d
Latency rose from 322ms to 412ms p95. Agents may start retrying or downgrading. Likely cache invalidation drift.
Action: Review Cloudflare cache rules + supplier-fanout timeout
Impact: Protect agent conversion
P2Customerconfidence 74%
Booking Engine SaaS partner trending down 18% MoM
Last 30 days vs prior 30. No deploys on partner side; no support tickets. Probably integration drift.
Action: CS to schedule a check-in this week
Impact: Defend €240k/quarter at risk
P3Contentconfidence 81%
187 hotels in Casablanca under 50% content completeness
Missing descriptions, photos, or facility lists. Search-result CTR is below segment median.
Action: Run sentiment_analysis batch + push enrichment task to Content team
Impact: +CTR uplift on MA segment
P3Operationsconfidence 83%
Refund spike on Dublin bookings W19-W20
Cancellation rate 14.2% (vs 8.1% prior 4w). Concentrated on 3 properties. Likely an overbooking issue.
Action: Cross-check supplier inventory vs sold rooms for the 3 properties
Impact: Process fix
Sample recommendations. In production: nightly LLM run over BigQuery marts. Ships to Slack + this dashboard with confidence + impact estimate per row.