Problem 2
Find the time, name who owes what, and start the follow-up — not only transcribe the meeting.
Professionals spend an average of 10-15 hours a week in meetings, and even more time coordinating them.
Finding a time that works for everyone, taking notes, identifying who needs to do what, and following up are tedious, manual tasks that drain productivity.
Existing solutions simply transcribe meeting audio or require users to manually click through calendar availability. They do not actively "think" about the user's schedule, negotiate with other participants, or enforce the action items discussed in the meeting. The challenge is not just to summarize text, but to create an intelligent system that acts as a proactive Chief of Staff.
This problem is relevant to organizations and teams including:
Build an AI-powered agentic system that manages the entire lifecycle of a meeting—from scheduling and negotiating times to extracting structured action items and initiating follow-ups.
Information Extraction: Accurately identify decisions, deadlines, and task owners from raw conversation.
Action-Oriented: Ability to trigger external actions (e.g., creating a calendar invite, sending an email).
Memory & Context: Remember what was discussed in past meetings to provide context for future ones.
The following levels are intended as guidance for increasing complexity. Participants may choose one or more levels based on their approach.
Level 1 (Basic Agency): An agent that takes a plain text meeting transcript, identifies action items, and generates a structured summary.
Level 2 (Tool Integration): An agent that connects to a simulated Calendar API. Given a prompt like "Schedule a 30-min sync with John next week," it checks availability, resolves conflicts, and creates the event.
Level 3 (Multi-Agent Collaboration): Two user-representing agents negotiating. (e.g., User A's Agent emails User B's Agent to find a time to meet. The agents talk to each other to find a mutually free slot without human intervention).
Level 4 (Advanced Multimodal & Proactive): The agent processes an actual audio recording (Speech-to-Text), detects different speakers, extracts tasks, and proactively drafts follow-up emails to each speaker containing their specific to-do lists.
A system that successfully demonstrates context extraction and autonomous scheduling/follow-up based on human inputs.
Participants are free to decide:
Which programming languages, frameworks (LangChain, CrewAI, AutoGen, LlamaIndex), or APIs (OpenAI, Gemini, HuggingFace) to use.
How to mock data, databases, and external tools.
The system architecture (single script, full-stack app, terminal-based).
1. Meaningful AI Usage
Do not just build a standard software app. The solution must rely on AI decision-making.
2. Autonomy over Hardcoding
The agent should figure out how to solve the problem, rather than relying entirely on hard-coded if/else statements.
3. Innovation Freedom
Participants are not restricted to specific tech stacks. Using smaller open-source models creatively can score just as high as using paid APIs.
What It Measures
Agentic Autonomy How well the system operates without human hand-holding
Whether the agent uses external tools correctly without crashing
Whether the AI self-corrects when it hits an obstacle
Scalability & Vision Whether the prototype could realistically be expanded into a real-world product
Ability to clearly demonstrate a working scenario within the 10-hour limit