Op Ed: Martijn Van Der Voort On How Legacy Tech Limits The Potential Of AI

Consultant Martijn van der Voort of AstraNomad pulls few punches in his prolific commentary on LinkedIn and elsewhere about the corporate travel space. Here, the longtime CWT tech leader, who left the travel management company about a year ago, challenges the industry to face the consequences of its surface fixes.

Orchestration is an emerging core concept for AI design and deployment, but in travel, it is inhibited by legacy infrastructure. This demands new architecture.

What do I mean by orchestration?

Agentic AI systems will do more than assist with tasks. They will orchestrate entire journeys, adapting in real time, aligning with policy and supporting the traveler throughout.

The promise is extraordinary. A traveler says, “Book me to New York next week, lowest carbon route and keep me in policy,” and the agent takes it from there: sourcing, booking, paying, abiding by policy, integrating into expense systems and notifying duty of care platforms.

Orchestration in an AI context refers to the real-time coordination of multiple intelligent agents, each capable of perceiving, reasoning and acting autonomously across a dynamic ecosystem of services and data. It’s not just about automating tasks but enabling agent-to-agent collaboration powered by large language models, secure APIs and event-driven architectures to handle complex, multi-step scenarios. This means a single agent can book travel, sync with policy engines, manage expenses and even extend operations to restaurants, events or gifts — basically, anything bookable or purchasable. Companies like Dataiera exemplify this next-gen model.

To realize this vision, we must acknowledge that the infrastructure supporting corporate travel today is not ready for this level of intelligence and autonomy.

The industry continues to depend on legacy systems. GDSs operate on older protocols. EDIFACT needs to be replaced by modern, API-based NDC standards and direct connects through aggregators or airline tech stacks. Many TMCs use platforms not designed for modularity or seamless system interaction. Many long-standing platforms (especially post-acquisition patchworks) lack real-time orchestration, open APIs or scalability. Systems still dependent on manual PNR stitching, mid-office scripts and batch-processing back offices are especially brittle. Legacy mid-office automation (e.g., scripts, queues, often still churning GDS PNRs), back-office reconciliation (e.g., fare auditing, invoice stitching) and HRIS/OBT/card integrations aren’t real-time or scalable. OBTs remain largely form-based.

Agentic AI, API-first, event-driven architectures don’t exist yet in corporate travel, but some players are closer to it than others.

Currently, AI in corporate travel is being layered over legacy systems. Much of the industry still treats AI as a surface enhancement, not a signal to rebuild what lies beneath. Core workflows remain the same. Traveler requests are processed, policies checked and bookings completed using intermediary systems that bridge old and new technologies. These systems may function for now, but over time they constrain value. The traveler may benefit from a better experience, yet the underlying systems remain brittle and hard to scale.

The future is not smarter use of current tools. It is about preparing for a reorientation over the next five to 10 years — a paradigm shift requiring long-term vision and investment in a new foundation to support autonomous systems. Agentic AI changes where intelligence resides, from human-driven workflows to agents acting on our behalf. Realizing this requires more than front-end upgrades; it demands a fundamental rethinking of architecture.

This is not disruption for disruption’s sake. It’s a call to lead. TMCs and their tech partners have an opportunity to replace legacy systems with modular, service-based infrastructure.

Other industries — logistics, finance, commerce — have already evolved toward automation-ready architecture. Corporate travel must follow, not to chase hype but to stay relevant in how value is delivered.

To adopt agentic AI responsibly, organizations need to launch cross-functional initiatives that address governance, ethics and readiness. If agentic AI is not yet on your roadmap, now is the time to ask: When will you start? Who will lead it? How will you build a governance framework around data privacy and performance?

Equally important is having a roadmap. Are you in the ideation, prototype, pilot or production phase? What use cases are priorities? The key is defining when and how autonomous agents will enter live environments, with appropriate oversight.

This is not just technical; it is a business model problem. Innovation requires aligning incentives to reward automation. TMC pricing often depends on high-touch support, which becomes more profitable when the tech breaks. Buyers are basically paying for headcount when tech goes awry, or when tech debt is not addressed. TMCs are not eager to change because they know they are clinging to business models others can do better in future. Time is running out.

* * *

The transition will take work, investment, discipline and, most importantly, a willingness to question the status quo. The outcome is worth it: A travel ecosystem that is truly intelligent, responsive and ready for what’s next.

Many are defensive, with borrowed language of vision and innovation, as they cling to outdated tech and fading models. But the future of corporate travel won’t be built on nostalgia. It will belong to those with the clarity to see and the courage to break free.

Agentic AI is not science fiction. It is approaching. It will come from building the foundations that those tools require to succeed. The question is not whether it will change our industry but who is prepared to shape that change.

This Op Ed was created in collaboration with The Company Dime’s Editorial Board of travel managers.