The direct-channel agent, or AI Concierge, has received less attention. While third-party platforms have dominated the conversation, branded AI concierge solutions such as Maya AI and Via.ai (wearevia.ai) now enable agentic commerce directly on travel providers' websites and digital channels.

​

Airlines have not adopted these platforms as extensively as third-party channels. For example, Via.ai, though well-suited as an agentic concierge for airlines, has primarily operated in the third-party retail channel. Notably, Via.ai has powered Issta, Israel's largest travel retailer, for the past 18 months. This deployment is now a documented production sales channel.

​

According to Via.ai's 2026 customer case study, Issta's AI sales agent, "Daniel," deployed across WhatsApp and Issta's website, generates $1.7 million in monthly gross booking value (June 2026), a $20.4 million annualized run-rate, across more than 10,000 conversations per month. On Issta's approximately $800 million in annual bookings, Daniel is already a fast-growing, measurable AI sales layer - in Via.ai's words, "not a pilot or a service-deflection layer."

​

The Issta deployment is strategically instructive due to its two-flow commercial model, with one AI layer serving two distinct sales engines:

​

Flow 1:  the channel that closes itself. For routine demand (flights, point bookings, standard packages), Daniel handles the journey end-to-end: clarifying preferences through natural conversation, presenting options, answering follow-ups, and closing the booking online. While conventional travel websites expect customers to arrive knowing exactly what they want, only to lose them to rigid search filters and abandonment, Daniel lets customers start with uncertainty and converts it.

​

The reported results: 5× conversion versus other digital lead sources, +30% average booking size, and roughly $700K per month in self-service sales. As Issta senior executive Tali Noy puts it: "Daniel has become a real commercial channel for us, not just a service layer. Customers ask, clarify, compare, and then book - and that changes the economics of the digital channel."

​

Flow 2:  the human team multiplier. For complex, higher-value demand multi-destination trips, cruises, and premium holidays, Daniel does not force customers into self-service. It qualifies the opportunity, captures preferences, budget signals, constraints, and purchase intent, and routes it to the right human specialist with the discovery work already done.

​

The economics of this flow are striking: specialists close 13% of Daniel-qualified leads versus a 10% baseline, at an average basket of $11,428 versus $3,125 / 4.75× the sales per lead of other sources generating roughly $1 million per month in specialist-closed sales. "The biggest impact is not that AI replaced people," notes Issta senior executive Nevo Gal. "It is that our specialists receive better-qualified opportunities with clearer customer intent  that allows the team to spend more time closing the right deals."

​

The direct approach maintains the provider's brand and customer relationship. To advance adoption, executives should assess organizational readiness for AI-driven interactions, map integration points with existing systems, and identify priority use cases. The Issta model suggests the first decision is not whether to deploy, but which flow to implement first: closing simple demand, qualifying complex demand, or turning every conversation into structured commercial intelligence.

​

Airlines are delaying full autonomy in booking not due to technological uncertainty, but because agentic AI conflicts with their legacy infrastructure, risk aversion, and internal silos:

​

• The Legacy Infrastructure Trap: True agentic commerce requires flexible modifications and pricing. However, airlines remain tied to rigid, PSS/PNR-era systems. In these environments, a single AI error regarding fare rules or EU261/DOT liabilities could result in a costly revenue crisis.
   

• The Low-Stakes Pivot: To avoid transactional liability, carriers are limiting AI to low-risk customer service and discovery roles. This is evident in recent launches such as Malaysia Airlines' Mavis, Delta's Concierge, and Qatar Airways' Sama.
   

• Organizational Paralysis: Revenue management teams are reluctant to relinquish control of pricing and bundling to autonomous algorithms. Additionally, the "agentic channel" lacks a single corporate owner, resulting in fragmented pilot programs across IT, marketing, and customer care.
   

• The Strategic Risk: Executives are misinterpreting modest initial consumer demand as a reason to delay action. This is a critical error. Early user behavior is already being used to train third-party AI models. If airlines use AI only for customer service while third-party platforms perfect autonomous booking, they risk losing the customer relationship entirely.
   

The Issta case demonstrates that a travel retailer with simpler infrastructure has already built a $20 million run-rate agentic sales channel, proving the technology is commercially viable. Carriers should deploy branded AI concierges on existing APIs now to capture customer intent before customer habits shift elsewhere.