As the landscape of global travel becomes increasingly complex, the role of artificial intelligence has shifted from a peripheral convenience to a core operational engine. Laurent Giraid, a seasoned technologist with deep expertise in machine learning and the ethical implementation of AI, offers a unique vantage point on this evolution. In this conversation, we explore the radical digital transformation at Omio, focusing on how the company moved beyond superficial automation to rebuild its entire engineering and consumer framework around generative models. We delve into the dramatic efficiency gains that have reduced months of work into weeks, the rise of conversational commerce that interprets traveler intent across thousands of providers, and the strict governance models that ensure human accountability remains at the center of innovation.
How does the transition to an AI-native enterprise fundamentally differ from simply integrating new software tools into existing workflows?
The shift to becoming an AI-native enterprise is a radical departure from the traditional “bolt-on” approach where companies simply add a layer of technology to their legacy systems. At Omio, this transition, spearheaded by CTO Tomas Vocetka, requires a complete ground-up redesign of how every internal function executes its tasks. Instead of just giving staff a new tool to use within their old habits, the company explicitly rejects superficial additions in favor of transforming the underlying operational execution frameworks. This means that processes are not just faster, but they are fundamentally reimagined to leverage generative capabilities as a baseline rather than an afterthought. It is a cultural and structural overhaul that forces the organization to move past outdated internal processes that were designed for a pre-AI world.
In what ways has the integration of OpenAI Codex reshaped the daily operations and responsibilities of an engineering team?
Integrating OpenAI Codex directly into the engineering operations has turned it into an essential component of the entire software development lifecycle, rather than just a coding assistant. Our engineers are now mandating its use for everything from preliminary research and architectural planning to active coding, automated testing, and even ongoing system maintenance. We have constructed custom internal connectors that link our proprietary data environments directly with these tools, which allows developers to bypass the tedious phases of basic information retrieval. By moving directly to active task execution within their integrated development environments, the team feels more like architects of solutions rather than just writers of syntax. This setup has matured so quickly that management is already expanding these practices into non-technical corporate functions to ensure the whole organization keeps pace with engineering’s new capabilities.
The reduction in development time at Omio is striking; how do these efficiency gains change the way a company handles experimentation and risk?
The data from our internal analysis is quite staggering, showing that the technical effort required to build specific products has plummeted to approximately 20 percent of what it used to be. We have seen projects that previously demanded a whole team of developers working for an entire fiscal quarter now being completed by a single engineer in roughly one month. This massive compression of delivery timelines has effectively lowered the cost and time barriers for software creation, which completely changes our appetite for risk. With faster cycle times, our engineering teams can afford to test experimental concepts and validate consumer demand with minimal resource expenditure. We can now use rapid prototyping to eliminate unviable features before we ever commit to full-scale production, allowing management to allocate capital and engineering hours with a level of precision that was simply impossible before.
How is the concept of conversational commerce transforming the way travelers interact with complex transportation networks?
Conversational commerce represents a departure from the legacy search-based interfaces that have frustrated travelers for decades, replacing them with a unified interface that understands natural language. In the past, a user wanting to get from Paris to Barcelona would have to navigate multiple websites and manually compare flights, trains, and buses, but our system now parses that consumer intent directly. By connecting OpenAI models to a proprietary inventory of over 3,000 transportation providers across 47 countries, we allow travelers to ask complex questions, like comparing multimodal routes from Rome to Florence, in plain English. The AI acts as the primary interface layer that mediates the interaction between the human and the vast, often fractured, global transportation network. It creates a seamless experience where the complexity of aggregating schedules and prices across ferries, buses, and flights is handled entirely behind the scenes by the model.
Given the risks of generative AI producing inaccurate information, how does a platform ensure that travel itineraries remain grounded in reality?
To solve the problem of “hallucinations” or outdated information, our architecture is specifically designed to ground every model response in live, verified data. The system analyzes the text input from the user and then pings our booking systems to pull real-time pricing and availability before constructing a travel path. This prevents the generative models from making suggestions based on static or outdated training data, which would obviously be a disaster in the fast-moving travel sector. By creating a dedicated experience that directly accesses the global transportation network we maintain, we ensure high-fidelity responses that are both highly personalized and bookable. It turns the AI from a generic advisor into a precise tool that provides journey options that are actually available for purchase at that exact moment.
While AI provides incredible acceleration, where do you draw the line regarding human accountability in a highly automated environment?
Our corporate policy is very clear: while generative tools are used as acceleration engines for development and analysis, human personnel retain full accountability for every line of code and every business outcome. As Tomas Vocetka emphasizes, AI helps us develop, analyze, and make decisions faster, but people stay in charge of the final results. This governance structure is a critical safeguard that prevents automated systems from independently executing any irreversible changes to our core booking infrastructure or routing algorithms. We have found that providing broad employee access to these tools, when paired with rigorous oversight models, creates an environment where we can prioritize speed without sacrificing systemic stability. It is about augmenting human expertise rather than replacing the critical judgment required to manage a global travel platform.
What is your forecast for the future of native generative customer experiences in the travel industry?
I believe we are heading toward a future where the very idea of “searching” for a trip will feel like an ancient relic, replaced entirely by interactions with intelligent systems connected to live networks. We will see a total shift where the interface layer is completely invisible, and the AI will proactively manage complex, multi-leg itineraries that adapt in real-time to delays or price changes without the user needing to lift a finger. The “fractured process” of the past will be replaced by high-fidelity, personalized journey planning that feels more like talking to a world-class travel agent who has every schedule in the world memorized. As these models become more grounded in proprietary, real-time data, the distinction between a “chatbot” and a “booking engine” will disappear, resulting in a single, fluid experience that prioritizes traveler intent over technical navigation.
