AIC Internal Competition 2026-2027: Agentic AI

This year's Internal Competition is all about building AI that can act, not just answer.

Your goal is to create an agentic AI project that can take a goal, figure out what needs to happen, use tools or outside information, and carry out multiple steps toward completing it. 

That could mean an agent that researches a topic, organizes your schedule, helps you study, automates a repetitive task, works with other agents, interacts with APIs, or does something completely ridiculuos but technically impressive. The opportunities are endless.

There is no single required problem to solve. You choose what you want to build.

What counts as "agenitc"?

Your project should go beyond a normal chatbot and demonstrate some level of autonomy. That could include things like:

- planning a sequence of actions

- using APIs, tools, external services

- retrieving information before answering

- adapting when something fails

You do not need to include all of these. Build something that makes sense for your idea. 
Think about "The agent loop" : Goal --> decide --> use tool(s) --> observe --> finish

Teams: You can compete either solo or in a team of up to 3 people. 

If you want teammates, use the #find-a-team channel in the AIC Discord.

• Functionality (20 pts)
  How well does the project work? Are the core features functional, reliable, and demonstrated successfully?
• Agentic Concepts (20 pts)
  How effectively does the project use agentic AI concepts such as reasoning, planning, tool use, memory, autonomy, or multi-step workflows?
• Creativity (20 pts)
  How original, interesting, or innovative is the idea and its implementation?
• Devpost Submission (25 pts)
  How clearly does the team explain the project, its purpose, how it works, and demonstrate it?
• Usefulness (15 pts)
  Does the project provide clear value, whether practical, educational, creative, or entertaining?