Automating Guest Insights in Hospitality with our AI Agent Hackathon

Background

A leading hospitality organization, operating a large portfolio of properties, has always prided itself on delivering exceptional guest experiences. But as their operations grew, so did the complexity of managing daily reservations, recognizing loyal guests, and ensuring front desk teams had the right information at the right time.

In the age of AI and automation, they saw an opportunity to modernize these processes, making them faster, smarter, and more personal.

Enter our team: a group of solution architects, developers, and designers passionate about digital transformation. We were invited to participate in a one-day hackathon, tasked with building a proof-of-concept AI agent that could automate the flow of reservation data and empower staff to deliver VIP treatment to their most frequent guests.

Project

Our Goal

The hackathon project was ambitious. Build an AI agent that could process daily reservation data, identify VIP guests, and notify the front desk so these guests could be welcomed with extra care.

The vision was clear, but the path to get there was anything but straightforward. Our team kicked off the day with a brainstorming session. The customer provided several scenarios, and after discussion, we zeroed in on the challenge of automating the daily reservation report.

The goal was to create a system where, every morning, the front desk would receive a curated list of guests who deserved special attention, frequent visitors or loyalty program members.

The Challenge

As with any hackathon, time was of the essence. But the technical hurdles quickly made themselves known.

Data Format

Reservation data arrived via email as an Excel file, but it wasn’t formatted as a table, a requirement for automation tools like Power Automate and Copilot. This meant manual formatting before ingestion, adding complexity and risk. 

Integration

The solution needed to read emails from a shared mailbox, process attachments, and store information in a scalable database. Mapping Excel columns to Dataverse required careful attention to data types and field names. 

Technical Hurdles

We encountered issues with data type mismatches (e.g., strings vs. numbers), mailbox configuration, and flow triggers. For example, the “length of stay” field was a string in Excel but needed to be a number in Dataverse, causing the flow to fail until resolved. 

Scenario Selection

With multiple use cases on the table, we spent valuable time deciding which scenario to tackle. In hindsight, having a clear use case defined before the hackathon would have saved time and reduced uncertainty. 

Time Pressure

Hackathons are a race against the clock. Every minute spent troubleshooting was a minute not spent building. We had to make quick decisions, adapt on the fly, and keep the end goal in sight. 

Our Solution

We Built a Proof-of-Concept AI Agent in One-Day

Despite the challenges, our team delivered an initial agent with its framework in place.

• Automated Flow: We built a Power Automate flow that triggered when a reservation email arrived in the shared mailbox. The flow:
  • Saved the Excel file to SharePoint for backup and historical reference.
  • Checked that the file was formatted as a table, then read the data.
  • Added each reservation as a new row in Dataverse, ensuring all data types matched and mandatory fields were populated.
• Agent Integration: Using Copilot Studio, we connected the agent to the Dataverse table. The agent could answer queries like “List all loyalty numbers attached to 10 or more reservations in the last 90 days,” making it easy for staff to identify VIP guests. 
• Notification System: Staff could trigger a manual flow by typing commands like “alert front office” in the agent chat. The agent would prompt for an email address, then send a curated list of VIP guests to the specified address—ensuring the right people had the right information at the right time. 
• Scalability: By leveraging Dataverse instead of SharePoint, the solution could handle large volumes of data and was ready for future expansion. While Dataverse comes with additional licensing costs, its scalability and reliability made it the right choice. 

The Outcome

The hackathon delivered a working prototype that exceeded expectations:

• Efficiency: The front desk now receives timely, automated notifications about VIP guests, allowing them to personalize service and maintain high standards. 
• Scalability: The solution is built on a foundation that can grow with the organization’s needs, handling increasing volumes of data without missing a beat. 
• Empowerment: Staff can interact with the agent to get real-time answers, trigger notifications, and ensure no VIP guest goes unnoticed. 
• Proof of Concept: While not production-ready, the prototype demonstrated what’s possible and set the stage for future development. 

Lessons Learned

Every hackathon is a learning experience, and this one was no exception:

• Data Preparation is Key: Automations require well-structured data. Formatting Excel files as tables before ingestion is essential. Future solutions should automate this step or work with data sources that are natively compatible. 
• Technical Details Matter: Pay close attention to data types, mailbox configurations, and flow triggers. Small mismatches can break automations and cost valuable time. 
• Scenario Selection: Define the use case early. Hackathons move fast, and clarity saves precious time. Next time, we’ll ensure the scenario is locked in before the event starts. 
• Set Expectations: Communicate to stakeholders that hackathon solutions are prototypes, not production-ready products. Setting the right expectations avoids disappointment and ensures everyone is on the same page. 
• Collaboration: Quick problem-solving and teamwork are vital, especially when surprises pop up. The ability to adapt, troubleshoot, and support each other made all the difference. 

Looking Ahead

This AI Agent Hackathon was a testament to what’s possible when creativity, technical expertise, and a commitment to service come together. We built a solution that not only solved a real-world problem but also laid the groundwork for future innovation.

As this hospitality leader continues to evolve, we’re excited to see how this prototype grows and adapts to meet their needs.