The Challenge
Hashtagtravels had built a strong reputation as a Switzerland destination specialist in the Indian outbound travel market, with a curated portfolio of Swiss itineraries spanning Alpine rail journeys, ski packages, summer hiking circuits, and luxury chalet stays. Their distribution relied on a network of over 400 travel agents across India — primarily in Tier 1 and Tier 2 cities — who booked Switzerland packages through Hashtagtravels as their preferred consolidator.
The operational problem was acute: the entire booking workflow was manual, fragmented, and fundamentally unscalable. Agents communicated package requests via WhatsApp or email. The Hashtagtravels operations team manually checked Travelport GDS for flight availability, cross-referenced hotel availability through Hotelbeds, called Swiss hotel chains directly for boutique properties outside Hotelbeds coverage, and assembled itineraries in Word documents. Vouchers were generated in Excel, formatted manually, and emailed as PDFs. The average processing time for a Switzerland package booking was 38–45 minutes.
This manual pipeline created three compounding business problems. First, the throughput ceiling was hard — a team of 8 operations staff could process approximately 65 bookings per day under ideal conditions, which by 2024 was already creating a 2-day turnaround for peak season bookings. Second, margin leakage was significant: without real-time hotel pricing, the operations team used weekly rate sheets that were frequently stale, and agents were routinely quoted packages below the current supplier cost. Third, the lack of a self-service agent portal meant Hashtagtravels had no scalable path to growing their agent network beyond the operational team's capacity to manage relationships manually.
The company engaged SpYsR following a recommendation from another travel operator in SpYsR's client network. The brief was clear: build a B2B portal that agents could use entirely self-service, with live GDS and hotel supplier integration, automated voucher generation, and a dynamic package builder that calculated margins in real time.
Our Approach
The 5-month engagement was structured in three phases: infrastructure and API integration (weeks 1–8), portal application development (weeks 9–16), and agent onboarding with parallel production run (weeks 17–20).
The most technically complex element was the multi-source inventory integration. Switzerland itineraries required combining Travelport GDS flights (primarily from Indian gateways — Delhi, Mumbai, Bengaluru — to Zurich and Geneva), Hotelbeds hotel inventory, Swiss Federal Railways (SBB) rail pass availability via their API partner, and direct inventory from a curated set of boutique Swiss hotel partners who were connected via a custom XML feed maintained by Hashtagtravels' product team.
SpYsR's architecture decision was to build a unified inventory aggregation layer rather than exposing the four source APIs directly to the frontend application. This aggregation service, built on Node.js and deployed as AWS Lambda functions, normalised responses from all four sources into a single internal inventory schema, handled currency conversion (Travelport returns EUR/CHF, Hotelbeds returns USD, SBB API returns CHF), and applied Hashtagtravels' configured markup rules per product category. The aggregation layer cached hotel availability at a 15-minute TTL in MongoDB, while flight availability was always fetched live from Travelport with no caching.
For the frontend, the team chose React with a component architecture designed around the agent workflow — search, package assembly, review, confirmation — rather than a conventional e-commerce funnel. This distinction mattered because agents assemble packages iteratively, frequently substituting one hotel for another or extending the itinerary by additional nights, which required the UI to support non-linear state management across the booking session.
What We Built
The agent portal is a React single-page application serving authenticated travel agents with role-based access. Agents can search and book six product categories: flights (via Travelport), hotels (via Hotelbeds + direct), rail passes (via SBB API), activity packages (curated Hashtagtravels catalogue), transfers, and visa assistance. The portal is accessible on both desktop and tablet, with a responsive layout tested against the device mix reported by Hashtagtravels' agent network.
The dynamic package builder is the platform's flagship capability. Agents construct Switzerland packages by combining products from any of the six categories, with real-time margin calculation displayed throughout the assembly process. The system applies Hashtagtravels' configured markup rules — which vary by category, season, and agent tier — and shows the agent the final sale price and the margin percentage simultaneously. This transparency enables agents to make informed commercial decisions on the spot rather than calling the operations team for pricing guidance. The package builder supports saving in-progress packages as drafts, which agents can return to and modify before confirmation.
Automated voucher generation replaced the manual Excel/PDF process entirely. On booking confirmation, the system generates a property-branded PDF voucher package — covering hotel confirmation, rail pass details, transfer confirmation, and itinerary summary — within 15 seconds of payment. The voucher engine is built on Node.js with a PDF generation library, using Handlebars templates maintained by the Hashtagtravels product team without engineering involvement. In the first month of operation, the system generated 2,340 vouchers automatically — a volume that would have required the equivalent of a full additional operations headcount under the previous process.
The agent management console — accessible to Hashtagtravels' internal team — provides real-time dashboards of agent activity, booking volumes, margin analytics by agent and package type, and a CRM-adjacent agent profile system tracking booking history and performance tiers. The console also manages the agent onboarding workflow, including document verification and access provisioning.
AWS Lambda functions with SQS-backed queuing handle the asynchronous operations — voucher generation, email dispatch, supplier confirmations — ensuring that booking confirmation response times remain under 3 seconds regardless of backend processing load.
The Impact
The platform went live in July 2025 after a 6-week parallel run period during which the top 40 agents from Hashtagtravels' network were invited to book through the portal while the operations team processed the same bookings through the legacy workflow. This parallel run validated accuracy, identified 12 edge cases in the SBB rail pass API integration that were resolved before full rollout, and built agent familiarity before the transition.
Agent onboarding reached 400 active users within 8 weeks of full public launch — well ahead of the 6-month target set in the business case. The self-service onboarding flow, which includes document upload, identity verification against GST records, and automated access provisioning, processes new agent applications in under 24 hours.
Average booking completion time fell from 38 minutes to under 10 minutes — a 74% reduction — measured across the first 1,000 portal bookings. The reduction is driven almost entirely by the elimination of manual supplier lookups and the voucher generation automation, rather than UI speed improvements.
Package margin improved by 31 percentage points against the prior year's same-period average. The primary driver is real-time pricing: the dynamic package builder applies current supplier rates at the moment of assembly, eliminating the stale rate sheet problem that had been causing systematic underquoting on hotel-heavy packages.
The platform operated at 99.97% uptime through its first quarter of live operation, supported by a multi-AZ AWS deployment and automated failover configured at the Lambda and MongoDB Atlas tiers.
