Wenrix RUNNER-UP Travel Innovator of the Year - #Phocuswright Innovation: Launch 2025
Summary
Amir Balaish, CEO of Wenrix, presents DeepFlow — an AI execution layer for post-booking travel servicing — as a runner-up in the Phocuswright Innovation Launch 2025 competition. Balaish opens by framing the industry's current predicament: every board meeting now demands answers on AI adoption, yet most companies have defaulted to deploying chatbots as a quick, visible response. His central argument is that chatbots are insufficient because they lack execution capability, and the real bottleneck in travel is not conversation but automated action.
Balaish cites a striking data point: the industry has automated roughly 40% of simple in-flight ticket servicing tasks, but the remaining 60% — the complex edge cases — consumes 82% of agent time. These edge cases involve fragmented systems, unstructured data, and airline-specific policies. Rather than a new problem, this is a long-acknowledged structural failure that has persisted because the infrastructure to fix it did not exist. Cheap offshore labor has historically been used to fill this gap.
DeepFlow is positioned as the infrastructure layer that closes what Balaish calls "the execution gap" — the distance between a traveler's service request and the supplier's action. The platform automates the entire post-booking lifecycle: refunds, exchanges, no-shows, flight disruptions, ancillaries, and more, across GDS (Edifact), NDC, and soon LCC channels.
The product is built on three core technology layers: (1) context understanding — interpreting the full service request and booking context including agency-specific business logic; (2) airline policy modeling — AI-determined penalties, fees, and workflows drawn from multiple data sources; and (3) automated execution with real-time sync to mid and back-office systems. Balaish emphasizes a deliberate strategic choice to go deep rather than wide: focusing exclusively on flights and on the back office rather than the front office.
DeepFlow is trained on over 50 billion real-world servicing data points and agent actions, integrated with major OTAs and TMCs, using LLM and pattern segmentation technologies. The platform auto-detects and learns from real-world outcomes. Eight years of persistent edge-case implementation underpin the system's current capability.
Proven results include 93% automation with one travel company — more than double the stated industry average — and over 90% accuracy on refunds and exchanges for CWT for a specific usage. Integration time is cited as one month, with a modular deployment option or connection via an MCP server.
Beyond cost reduction, Balaish positions DeepFlow as a revenue enabler: proactive disruption handling, upsell opportunities, and real-time fare policy delivery to search — shifting servicing from a cost center to a revenue driver. The Q&A covers pricing (per-transaction, complexity-tiered, pay-as-you-go), ROI justification (transparent cost comparison against manual handling), edge-case prioritization (now largely solved after eight years), and the evolving buyer profile (still primarily operations teams, but with growing broader management awareness).
Key Takeaways
- •The travel industry has automated ~40% of simple servicing tasks, but the complex 60% still consumes 82% of agent time — this is where DeepFlow focuses.
- •Chatbots without an execution layer are insufficient; the real problem in travel is automated action, not conversation.
- •DeepFlow achieves 93% automation on post-booking servicing for one major travel company, more than double the stated industry average.
- •Eight years of persistent edge-case implementation is a structural moat that cannot be replicated quickly by new market entrants.
- •Going deep on a narrow vertical (flights only, back office only) produces better outcomes than broad horizontal coverage.
- •Integration is designed to take one month and can be deployed one module at a time, lowering adoption friction.
- •Pricing is per-transaction and complexity-tiered (pay-as-you-go), making ROI immediately calculable by comparing against manual handling costs.
- •DeepFlow supports an MCP server interface so that front-office chatbots can delegate execution tasks directly to the platform.
- •The platform enables a shift from reactive servicing to proactive value generation — real-time disruption handling and upsell — turning operations into a revenue driver.
- •Buyer awareness is expanding beyond operations teams to broader management as automation becomes a strategic organizational priority.
Transcript
Hi everyone. Since the introduction of AI, every board meeting includes two key questions. How are you using AI to improve servicing? How are you using AI to make your company more efficient? And with headlines screaming, AI replaces thousands of jobs. Everyone is under pressure to act fast. So they land chat bots. Quick, visible, easy headlines. And while chatbots are great, they have one major drawback. They cannot necessarily execute. Because the real bottleneck in travel isn't about conversa...
