We built a 5-component AI operations workflow over 11 weeks. (1) Revenue Intelligence Engine: automating rate decisions across all properties and channels. The engine: forecasting demand by property using: historical booking patterns (3 years of data by property, room type, and source market), forward-looking signals (Barcelona flight search volume from key markets — UK, France, Germany, US — event calendar impact estimates, cruise ship schedule for Barcelona port), competitor pricing (automated rate monitoring for 5-8 competitor properties per location, refreshed every 6 hours), and weather and seasonal patterns. Setting optimal rates by property, room type, date, and channel using: demand forecast, price elasticity models (trained on each property's historical booking response to rate changes), channel-specific strategies (higher rates on high-commission channels, best-rate guarantee on direct channels, contracted rates for tour operators), and revenue management constraints (minimum rate floors ensuring profitability, maximum rate caps per brand positioning).
Updating rates across all 18 channels automatically via channel manager API integration. Rate update frequency: every 6 hours during peak season (vs. 2-3 times weekly manually). (2) Guest Communication Engine: automating multilingual guest communications. The engine: generating pre-arrival communications (booking confirmations, pre-arrival information — transport options, check-in procedures, local recommendations — in the guest's preferred language), handling in-stay requests (restaurant reservations, tour bookings, special requests — triaging by complexity and routing simple requests to automated fulfilment, complex requests to concierge staff), responding to post-departure reviews (acknowledging positive reviews, addressing negative reviews with property-specific responses — drafted by AI, reviewed by guest relations manager before posting), and managing complaint resolution workflows (categorising complaints, routing to appropriate department, tracking resolution, generating follow-up communications). Guest communication processing: 85% of routine communications handled automatically (confirmations, standard information, simple requests). Response time for guest inquiries: reduced to 25 minutes average (from 4.2 hours).
Multilingual capability: Catalan, Spanish, English, French, German, Italian — with cultural nuance appropriate to each source market. (3) Compliance Automation Engine: automating regulatory documentation and filing. The engine: calculating IEET tourist tax for every guest stay (applying correct rate based on property star category, municipality, guest age — children under 16 exempt, and duration — maximum 7 nights taxable), generating quarterly IEET declarations for filing with Agència Tributària de Catalunya, maintaining tourism license compliance records (tracking registration status, renewal dates, and inspection readiness for all 22 properties), automating HACCP documentation for F&B operations (daily temperature logs from connected sensors, cleaning schedule verification, supplier documentation tracking), and generating SII (Suministro Inmediato de Información) real-time VAT reporting to AEAT. IEET tax filing: fully automated — zero errors (vs. 8 corrections in the previous year). HACCP documentation: 90% automated through sensor integration and scheduled workflow prompts.
(4) Workforce Management Engine: automating staff scheduling across all properties. The engine: forecasting staffing requirements by property and department based on occupancy forecasts, revenue forecasts, and event schedules, generating optimal schedules considering: Spanish labour law constraints (maximum working hours, mandatory rest periods, overtime limits per Estatuto de los Trabajadores), collective bargaining agreement provisions (convenio colectivo — specific to Catalan hospitality), employee preferences and availability (captured through a mobile app), skill requirements (front desk staff need language skills matching expected guest mix), managing seasonal contract administration (generating temporal contracts per Spanish labour law, tracking contract durations and renewal/termination dates), and producing labour compliance documentation (jornada laboral records — mandatory time tracking under Spanish law). Scheduling time: reduced from 120 hours monthly to 20 hours (manager review and adjustment of AI-generated schedules). Labour law compliance: zero violations (vs. 4 in the previous year).
(5) Financial Operations Engine: automating invoice and payment processing. The engine: receiving invoices via email and supplier portals, extracting key data (supplier, invoice number, date, line items, amounts, VAT — validating Spanish factura requirements), matching invoices against purchase orders and delivery notes (albaranes), routing for approval based on amount and cost centre (property manager approval for standard invoices, operations director for amounts above threshold), coding to correct property, department, and general ledger account, validating VAT treatment and generating SII entries, and scheduling payments to optimise cash flow while capturing early payment discounts. Invoice processing time: 3 minutes per invoice (from 18 minutes). Payment errors: reduced from 2.8% to 0.3%. Early payment discount capture: increased by 40%.