Conversations That Matter

A conversation with Rohit Gupta and Shiv Kapur of AI Concierge on AI-powered travel, human oversight, privacy, and the future of intelligent agents 

At The Talk Tree, we were intrigued by AI Concierge and its vision of creating a more intelligent and seamless way to plan and manage travel.

What caught our attention, however, was not simply the technology behind the platform. It was the question at the heart of its design: How much should we allow AI to do on our behalf, and where should humans remain firmly in control?

We spoke with Rohit Gupta, CEO of AI Concierge, and Shiv Kapur, professional golfer and AI Concierge board member, about the experiences that shaped the platform, the challenges of modern travel, and the ethical questions emerging as AI systems become increasingly capable of acting on our behalf.

Their perspectives are shaped by very different professional experiences. Gupta brings an entrepreneurial and business perspective, while Kapur's more than two decades of international professional golf have made travel an integral part of his career.

Together with the other members of the AI Concierge leadership team, including Anish Kushalapa, who leads product and technology, and Dr. Hassan Al Sayegh, who guides strategic partnerships and ecosystem development, those experiences have informed a platform built around a central idea: travel should feel like one connected journey rather than a collection of disconnected bookings.

But AI Concierge's approach raises a broader question that extends well beyond travel: As AI moves from providing information to making recommendations and taking action, how much authority should we give it?

In this conversation, we explore AI-powered travel, human oversight, privacy, transparency, AI agents, and the boundaries that AI Concierge believes should remain in place.

A Platform Built Around Continuity

1. Let's start with the story behind AI Concierge. What made both of you believe there was an opportunity to rethink the way people plan and manage travel?

AI Concierge: From the outset, the company was guided by a simple conviction: travel works best when everything moves together.

Instead of treating flights, hotels, rail and transfers as separate transactions, AI Concierge was designed as one continuous trip. Bookings live inside a single plan rather than across disconnected confirmations. When plans change, the trip adapts rather than resets. Totals are visible before commitment, and personal documents can be stored securely and permissioned for trusted travelers without repeated collection.

The platform draws from global airline and hotel networks, including more than 750 airlines and over 2 million hotels, but its emphasis is not simply on breadth. It is on cohesion.

Intelligence runs quietly in the background, monitoring itineraries, learning preferences, and helping adjust when disruptions occur.

This focus on continuity reflects both the personal and professional experiences behind the company.

2. You come from very different professional backgrounds. Rohit, you've built and led businesses across different industries, while Shiv, you've spent much of your career traveling internationally as a professional athlete. How did those different experiences shape the idea behind AI Concierge?

AI Concierge: There are actually four of us who share the belief that existing travel platforms can be individually good while still being complicated for travelers because people have to use multiple platforms to manage one trip.

Our backgrounds are diverse, but that diversity has helped shape the platform.

Anish Kushalapa leads product and technology at AI Concierge, with a background spanning analytics, finance, and hospitality operations. His work focuses on designing systems that remain reliable under real-world travel conditions, where multiple stakeholders, changing plans, and high expectations around speed and clarity all come into play. At AI Concierge, Anish oversees product strategy, AI workflows, and global supplier integrations, owning the end-to-end booking and trip-management experience.

Dr. Hassan Al Sayegh brings decades of experience across the UAE's investment, tourism, and technology landscape. He has served in senior advisory capacities within national initiatives and holds board roles with major government investment entities. At AI Concierge, he guides high-level partnerships, ecosystem positioning, tourism collaborations, and long-term strategic alignment, helping the company build credibility and integrated value across the UAE and the wider region.

Our board is diverse, with Shiv bringing experience as a professional athlete with extensive international travel and Rohit bringing entrepreneurial experience combined with significant work and leisure travel.

All of our careers have unfolded in very different arenas, yet we have operated in environments where coordination, timing, and structure are non-negotiable. Those lived experiences influenced how AI Concierge was built.

3. Shiv, having spent so much of your life traveling around the world, what were some of the frustrations or challenges you experienced with travel that made you think, "There has to be a better way"?

Shiv Kapur: I wish there was something like AI Concierge 20 years ago!

My professional golf career has spanned more than two decades of international competition, and travel has never been incidental. It has been part of performance.

Weeks are mapped around flights, recovery windows, and tournament schedules. A poorly timed arrival can affect preparation, while fragmented logistics add mental strain in high-pressure environments.

That reality shaped how AI Concierge approaches individual travel.

The system treats the itinerary as a living object, something that stays intact from search through change. Preferences are remembered. Context is preserved. Adjustments cascade intelligently when one element shifts.

The aim is not to make travel louder or more complex, but to reduce unnecessary friction so attention can stay where it belongs.

From Search Engine to Travel Companion

4. AI Concierge goes beyond simply recommending destinations or hotels. What makes the platform feel more like a personal travel companion than a traditional travel search or booking service?

AI Concierge: The system treats the itinerary as a living object, something that stays intact from search through change.

Preferences are remembered. Context is preserved. Adjustments cascade intelligently when one element shifts.

The aim is not to make travel louder or more complex, but to reduce unnecessary friction so attention can stay where it belongs.

We have integrated with suppliers across hotels, flights, taxi and transfer services, eSIMs, airport services such as Meet & Greet, airport lounges and VIP boarding, European trains, and activities, with additional categories coming soon.

This allows us to ground AI suggestions in live supplier data and pricing. Rather than making recommendations based primarily on inference, the system can consider the user's preferences alongside available supplier inventory to make a more relevant match.

Within the next month, we are also launching a multi-service checkout flow. After the AI plans the trip based on the user's preferences, there will be one button to check out across multiple bookings and services.

This creates a more seamless purchasing process while preserving an important principle: the AI can assist with planning, but sensitive actions such as payment remain under the user's control.

The Line Between Assistance and Autonomy

5. One of the things we find fascinating about AI is its transition from giving us information to actually taking action for us. How do you think about that distinction at AI Concierge?

AI Concierge: At AI Concierge, we draw the line very clearly regarding what we want AI to do.

AI is a planning assistant, not your replacement.

Our AI agent helps with situations where there are too many options to evaluate manually.

What activities should I book? How can I get the most out of my budget? What hotels match my preferences? Which activities conflict with each other? If I am booking a trip for multiple people, how do I take everyone's interests into account?

These are the kinds of questions our AI is designed to answer.

However, we draw a clear line here. AI is useful for planning, but it should not be used where personal or sensitive information is unnecessarily exposed, nor should it simply be given the ability to book an entire trip independently.

We are here to simplify the planning process, not to orchestrate the booking flow end-to-end.

6. When an AI can help make bookings, change plans, and handle travel decisions, how much freedom should it have? Where do you believe the human traveler should always remain in control?

AI Concierge: The human element must be the final decision-maker, with complete veto power. At AI Concierge, we truly believe in that.

The AI can create plans and provide suggestions when changes occur. These suggestions are guided by the preferences a user specifies and their past bookings.

However, the AI cannot execute changes that materially affect travel bookings, such as payment, cancellation, or confirmed booking modifications.

We intentionally cut the system off from these functionalities because we believe individuals should maintain complete control over their data and decision-making.

Trust, Privacy and the "Blast Radius" of AI

7. Trust becomes especially important when an AI is dealing with someone's money, personal information, and travel plans. What do you think an AI system needs to do before people can genuinely trust it with those responsibilities?

AI Concierge: Given the unpredictability we have seen with frontier AI agents making decisions autonomously, sometimes against human requests, we find guardrails and barriers even more important when it comes to respecting user privacy and decision-making.

AI agents have access to user preferences, but they are completely separated from sensitive user information and personally identifiable information.

Furthermore, card information is stored with our payment provider, Stripe. We cannot access it even if we wanted to.

All payments require mandatory OTP verification sent via the user's bank, making it impossible for bookings to be made without the user's consent.

8. Travel information can reveal quite a lot about a person, including where they are going and when they will be away. How do you balance personalization and convenience with protecting people's privacy?

AI Concierge: User conversations are secured on our servers, where we enforce state-of-the-art firewalls, row-level security (RLS), and other security measures.

Our approach is based on giving the AI access to the preferences it needs to provide useful recommendations while keeping sensitive personal information separated from those AI workflows.

9. What happens when AI gets something wrong? How important is it for a traveler to understand why an AI made a particular recommendation or decision?

AI Concierge: There is always a chance that AI can get something wrong, which is why we limit its scope to suggestions and planning.

Transparency about recommendations is important, and we clearly state why an option was suggested based on the user's preferences and settings.

Planning is a high-value, low-risk task because the user can override or discard suggestions they are not satisfied with.

The potential impact when AI gets something wrong is therefore intentionally limited because we have separated the AI from critical booking flows.

10. AI Concierge also has applications in corporate travel, where budgets, policies, approvals, and business decisions can come into play. What additional ethical considerations do you think arise when AI begins operating within an organization?

AI Concierge: Our corporate platform, which is currently in development, works on the exact same principles as our B2C platform.

AI works by providing suggestions for planning.

Budgets, policies, approvals, and business decisions are not governed by AI. Instead, we use rules-based enforcement, as many existing non-AI products do.

AI is useful as an assistant; it should not be arbitrarily applied to everything.

Rules and enforcement should be structured and clear. There is no need for AI in every evaluation process.

Keeping Humans in the Loop

11. As AI agents become more capable, do you think people will eventually become comfortable giving them greater authority to make decisions and take actions on their behalf? Or do you think there will always be certain decisions where people will want to remain directly involved?

AI Concierge: We think one of the things that makes us intrinsically human is the ability to think and make decisions for ourselves.

Regardless of the autonomy and capability of AI, we should still have the ability to decide what works for us and what does not.

It is not possible to predict with absolute certainty what someone will want. Therefore, the need for a human in the loop will always remain.

Looking Ahead

12. Finally, when you look ahead, what do you hope AI Concierge becomes? And more broadly, what do you think the relationship between people and AI-powered agents will look like five years from now?

AI Concierge: Just like many of you, we are equally concerned by the unpredictability of frontier models making the headlines.

Governance, guardrails, and clear boundaries around what AI can access are essential to maintaining positive outcomes with AI.

Our hope for AI Concierge is to expand our supplier network and inventory, focusing on better rates for consumers and a wider selection. We intend to add rental cars and cruises to create a truly end-to-end travel platform.

This may come as a surprise because it is less about AI and more about traditional travel suppliers. However, when we drill down to our core belief that AI is a suggestion engine, our main leap will come from achieving a wider range of options to select from.

That wider range will allow us to provide exclusive bookings and better pricing while simplifying travel planning and research.

Ultimately, the goal is not simply to make AI more autonomous. It is to make travel easier to navigate while keeping meaningful decisions in human hands.

A Different Vision of the AI Agent

AI Concierge's approach reflects a broader debate taking place across the technology industry: Should increasingly capable AI systems be designed to act independently, or should their usefulness be measured partly by how effectively they preserve human agency?

For AI Concierge, the answer is relatively clear. AI can process options, understand preferences, identify conflicts, and offer recommendations. But when decisions involve money, sensitive information, cancellations, or meaningful changes to a person's plans, the company believes the final authority should remain with the human traveler.

That distinction may become increasingly important as AI agents move beyond answering questions and begin interacting with the systems that shape everyday life.

The future of AI may not ultimately be defined by how much autonomy we can give machines, but by how thoughtfully we decide where that autonomy belongs.