Inspiration
Large-scale sporting events such as the 2026 World Cup require thousands of operational tasks to be coordinated across teams, venues, vendors, and city officials. While modern project management platforms help track work, teams still spend significant time manually reviewing issues, identifying blockers, coordinating stakeholders, and ensuring readiness before match day.
We were inspired by a simple question: What if an AI agent could act as an operations coordinator instead of just a chatbot?
Rather than answering questions, we wanted to build an agent that could understand a goal, create a plan, gather information from project management systems, identify risks, and help teams take action. The World Cup provided an exciting real-world scenario where operational excellence directly impacts the experience of millions of fans.
This challenge aligned perfectly with the vision of agentic AI: moving beyond conversation and empowering users with systems that can reason, plan, and execute tasks under human supervision.
With MatchReady AI, we set out to demonstrate how Gemini, Google Cloud Agent Builder, and GitLab MCP can work together to transform event operations from reactive management into proactive coordination. The result is an intelligent operations copilot that helps organizers prepare for match day by automatically analyzing project status, detecting risks, assigning priorities, and generating actionable readiness reports.
What it does
MatchReady AI is an agent-powered operations copilot designed to help organizers prepare for major sporting events such as the 2026 World Cup.
Instead of simply answering questions, the agent takes a high-level goal—such as "Prepare for the Brazil vs. Spain match"—and autonomously plans the steps required to assess operational readiness.
Using Gemini's reasoning capabilities and GitLab's MCP integration, MatchReady AI can:
- Analyze project issues, milestones, and task boards in GitLab.
- Identify unresolved blockers that could impact match-day operations.
- Prioritize critical tasks based on urgency and dependencies.
- Generate readiness reports for organizers and stakeholders.
- Create follow-up tasks for missing requirements or operational gaps.
- Summarize risks and recommend mitigation actions.
- Track progress across multiple teams and workstreams.
- Provide a centralized operational view for venue managers and event coordinators.
The agent follows a human-in-the-loop workflow, ensuring that organizers remain in control of critical decisions while benefiting from AI-powered planning and automation.
By combining Gemini's advanced reasoning, Google Cloud Agent Builder's orchestration capabilities, and GitLab MCP's project management tools, MatchReady AI helps event teams reduce manual coordination effort, improve visibility, and ensure every match is ready for kickoff.
How we built it
MatchReady AI was built using Gemini, Google Cloud Agent Builder, and the GitLab MCP (Model Context Protocol) server to create a truly agentic workflow capable of planning, reasoning, and taking action.
Architecture
The system consists of four main components:
- Frontend Interface
- A web-based dashboard where event organizers can submit operational goals and review recommendations.
- Provides visibility into agent decisions, identified risks, and generated action plans.
- Google Cloud Agent Builder
- Acts as the orchestration layer for the agent.
- Manages conversation flow, tool calling, task planning, and execution workflows.
- Gemini
- Serves as the reasoning engine.
- Breaks down high-level objectives into actionable steps.
- Evaluates project status, identifies blockers, prioritizes tasks, and generates readiness reports.
- GitLab MCP Server
- Provides secure access to GitLab project data and actions.
- Enables the agent to retrieve issues, milestones, epics, merge requests, and project status information.
- Allows the agent to create follow-up tasks and update project workflows when approved by users.
Agent Workflow
When a user submits a request such as:
"Prepare the venue operations team for the upcoming match."
The agent:
- Retrieves relevant project data from GitLab through MCP.
- Analyzes open issues, milestones, and dependencies.
- Detects critical blockers and operational risks.
- Generates a readiness assessment.
- Recommends mitigation actions and task priorities.
- Creates or updates work items after user approval.
- Produces a comprehensive match-day readiness report.
Technologies Used
- Gemini for reasoning and planning
- Google Cloud Agent Builder for agent orchestration
- GitLab MCP Server for tool integration
- Python for backend services
- FastAPI for API endpoints
- React for the frontend dashboard
- Docker for containerized deployment
- GitLab APIs exposed through MCP tools
By combining these technologies, we created an agent that goes beyond traditional chat interactions and actively helps operations teams coordinate, plan, and execute complex event workflows.
Challenges we ran into
Building an agent that could reliably assist with real-world operations introduced several technical and design challenges.
Designing True Agentic Behavior
One of the biggest challenges was moving beyond a traditional chatbot experience. We wanted MatchReady AI to do more than answer questions—it needed to understand goals, create execution plans, gather information from multiple sources, and recommend actions. Designing workflows that balanced autonomy with user control required careful planning and testing.
Integrating GitLab Through MCP
Working with the GitLab MCP server required us to rethink how tools are exposed to the agent. Instead of directly calling APIs, the agent needed to discover and use MCP-provided capabilities appropriately. Ensuring that the agent selected the correct tools and executed actions in the right sequence was a significant challenge.
Managing Context and Priorities
Operational projects often contain hundreds of issues, tasks, and dependencies. Determining which items were most important for match-day readiness required effective filtering, summarization, and prioritization strategies. We had to ensure the agent focused on high-impact risks without overwhelming users with unnecessary information.
Balancing Automation and Human Oversight
While automation can save time, event operations involve critical decisions that should remain under human control. We implemented a human-in-the-loop approach where the agent can recommend actions and prepare updates, but users approve important changes before execution.
Creating Meaningful Readiness Assessments
Another challenge was generating reports that were both concise and actionable. Early versions either produced too much information or lacked operational context. We refined our prompts and workflows so the agent could highlight the most critical blockers, risks, and next steps in a format that organizers could quickly understand.
Building for Real-World Scale
Large events involve multiple teams working simultaneously across different operational domains. Designing an architecture that could support numerous projects, tasks, and stakeholders while maintaining responsiveness required careful consideration of scalability and workflow management.
Despite these challenges, overcoming them helped us build a more practical and reliable agent that demonstrates how AI can assist with real operational coordination rather than simply providing information.
Accomplishments that we're proud of
One of our biggest accomplishments was successfully building an agent that goes beyond simple question-answering and actively helps users accomplish operational goals. MatchReady AI demonstrates how modern AI systems can reason, plan, and coordinate tasks while keeping humans in control of critical decisions.
Building a True Multi-Step Agent
Rather than creating a traditional chatbot, we developed an agent capable of:
- Understanding high-level objectives.
- Breaking goals into actionable tasks.
- Gathering information from GitLab through MCP.
- Identifying blockers and operational risks.
- Generating actionable recommendations.
- Supporting decision-making with readiness reports.
This end-to-end workflow showcases the practical value of agentic AI in real-world operations.
Successfully Integrating GitLab MCP
We are proud of our meaningful integration with the GitLab MCP server. The agent can interact with project data, analyze work items, track dependencies, and assist with project coordination through standardized MCP tools rather than relying on custom integrations.
Creating a Real-World Use Case
Many AI demos focus on theoretical capabilities. We chose to solve a practical challenge inspired by large-scale event operations, where coordination, visibility, and risk management are critical. This allowed us to demonstrate how AI agents can deliver measurable value in realistic scenarios.
Human-in-the-Loop Design
We designed the system to keep users in control. The agent provides recommendations, prioritizes tasks, and prepares actions, while final decisions remain with event organizers. This balance between automation and oversight makes the solution more trustworthy and practical for real-world adoption.
Delivering Actionable Insights
We are particularly proud of the agent's ability to transform large volumes of project information into concise operational intelligence. Instead of requiring users to manually review numerous issues and tasks, MatchReady AI surfaces critical blockers, highlights risks, and presents clear next steps.
Demonstrating the Future of AI Agents
Most importantly, we successfully demonstrated how Gemini, Google Cloud Agent Builder, and GitLab MCP can work together to create an intelligent operations copilot. The project showcases a future where AI agents do not simply answer questions—they help teams plan, coordinate, and execute complex workflows more efficiently.
What we learned
Building MatchReady AI provided valuable insights into the emerging world of agentic AI systems and how they differ from traditional chatbot applications.
Agent Design Is More Important Than Model Capability
One of our biggest lessons was that building an effective agent is not just about using a powerful language model. The quality of the workflow, planning logic, tool integration, and decision-making process has a significant impact on the final user experience. Carefully designing how the agent reasons through tasks proved just as important as the model itself.
MCP Simplifies Tool Integration
Working with the Model Context Protocol (MCP) demonstrated the value of a standardized approach to connecting AI systems with external tools. Instead of building custom integrations for every service, MCP provides a structured way for agents to discover and interact with capabilities, making systems more modular and extensible.
Human Oversight Remains Essential
We learned that fully autonomous execution is not always the best solution for operational workflows. Users want transparency and control over important decisions. Incorporating a human-in-the-loop approach improved trust, reduced risk, and made the system more practical for real-world use.
Context Management Is Critical
Operational projects generate large amounts of information. One of the key lessons was that providing more data does not necessarily create better outcomes. Effective summarization, prioritization, and context selection are essential for helping users focus on what truly matters.
Multi-Step Reasoning Creates Real Value
The most impactful experiences came when the agent could combine multiple actions into a coherent workflow. Rather than answering isolated questions, the agent could gather information, analyze dependencies, identify risks, and propose solutions. This reinforced our belief that the future of AI lies in task completion rather than conversation alone.
Building for Real-World Problems Changes Everything
Developing a solution around a realistic event-operations scenario forced us to think beyond technical implementation and focus on user needs. We learned the importance of reliability, explainability, and actionable outcomes when designing AI systems intended for operational environments.
Overall, this project deepened our understanding of how Gemini, Google Cloud Agent Builder, and GitLab MCP can be combined to create practical AI agents that help people accomplish meaningful work, not just retrieve information.
What's next for MatchOps AI
While MatchOps AI already demonstrates a functional agent that can analyze GitLab projects and support event readiness, there are several directions we are excited to explore next.
Expanding Beyond GitLab
We plan to extend the agent to integrate with additional operational systems such as incident management, communication platforms, and scheduling tools. This would allow MatchOps AI to coordinate across the full lifecycle of event operations rather than being limited to project tracking.
Real-Time Event Monitoring
A major next step is enabling real-time monitoring of operational signals during live events. By continuously ingesting updates, the agent could detect emerging issues, surface anomalies, and recommend immediate actions to reduce downtime or disruption.
More Advanced Planning Capabilities
We aim to enhance the agent’s planning abilities so it can handle longer-horizon goals such as multi-match scheduling, resource allocation, and contingency planning. This includes improving how the agent simulates dependencies and evaluates risks across complex event timelines.
Deeper Human-AI Collaboration
Future versions will further refine the human-in-the-loop experience, making it easier for users to review, approve, and modify agent-generated plans. We also plan to introduce more transparent reasoning outputs so users can better understand why specific recommendations are made.
Domain Expansion Beyond Sports
Although MatchOps AI is inspired by World Cup operations, the underlying system is applicable to many domains. We plan to adapt it for enterprise project management, large-scale conferences, and other coordination-heavy environments where multi-team collaboration is critical.
Smarter Learning From Historical Data
We also aim to incorporate historical project data to help the agent learn patterns over time. This would enable better prediction of risks, improved prioritization, and more accurate readiness assessments based on past outcomes.
Overall, our goal is to evolve MatchOps AI into a general-purpose operations agent that can help teams plan, coordinate, and execute complex real-world workflows with greater efficiency and confidence.
Log in or sign up for Devpost to join the conversation.