CampusFlow AI
Inspiration
Campus life involves many repetitive coordination tasks—collecting requests, following up with students, coordinating with staff, tracking pending work, and sending reminders. These tasks may seem simple, but they consume time and require people to repeatedly check messages, spreadsheets, and forms.
As a student, I wanted to explore a different approach: What if an AI agent could take responsibility for these repetitive coordination tasks instead of simply answering questions?
This led me to build CampusFlow AI, an autonomous AI agent designed to act as a digital coordination assistant for campus activities.
What the Project Does
CampusFlow AI helps automate repetitive campus coordination workflows.
Instead of requiring a person to manually manage every request, the agent can:
- Understand incoming requests
- Identify what action is required
- Organize and prioritize tasks
- Coordinate information between relevant people
- Track the status of tasks
- Send reminders and updates
- Report completed and pending tasks
The goal is not to replace people, but to reduce repetitive work so people can focus on decisions and activities that actually require human attention.
How I Built It
I built the project as a solo developer, using an agent-based architecture.
The core of the system uses the Strands Agents SDK to create the autonomous AI agent and its workflow. The agent receives a task, reasons about what needs to be done, uses the appropriate tools, maintains the task state, and produces an outcome.
The project architecture consists of:
User → CampusFlow Interface → AI Agent → Tools/Actions → Task State → Notifications/Updates
The agent is designed around the idea that an AI system should not just generate text—it should be able to take meaningful actions through tools.
I focused on creating a practical campus use case rather than building a general-purpose chatbot.
What I Learned
This project helped me understand the difference between a traditional chatbot and an AI agent.
I learned how agents can:
- Receive a goal rather than just a question.
- Break a task into smaller steps.
- Decide which tool or action is appropriate.
- Maintain the state of an ongoing task.
- Execute actions and verify their results.
- Communicate the outcome back to the user.
I also learned that building an agent is not only about the AI model. Tool design, workflow design, state management, error handling, and clear task boundaries are equally important.
Most importantly, I learned to think about AI from a human perspective: Where can an agent genuinely save someone time?
Challenges
One of the biggest challenges was designing the agent so that it could handle a workflow rather than simply respond with an answer.
Another challenge was deciding how much autonomy to give the agent. A useful agent needs enough freedom to complete tasks, while important actions should still be predictable and controllable.
I also had to think about handling incomplete information, keeping track of task status, and making the agent's actions understandable to the user.
Since I built the project alone, I also had to work across multiple areas—from designing the idea and architecture to implementing the agent and preparing the final prototype.
What's Next
CampusFlow AI is currently focused on campus coordination, but the same agent architecture could be extended to other repetitive administrative workflows.
Future versions could include more specialized agents, integrations with campus systems, calendar and email workflows, richer notifications, and human approval for sensitive actions.
The long-term vision is simple:
«AI should handle the repetitive coordination work, while humans stay in control of the decisions that matter.»
CampusFlow AI is my exploration of what that future could look like.
Log in or sign up for Devpost to join the conversation.