Note
This project is for demonstration purposes only and not intended for production use.
This repository contains an AI-agent-based tool assisting in scheduling meetings. It makes use of the AI agent framework Agno via Python.
The goal is to assist users with finding time slots for meetings given the following constraints:
- Each user has a calendar with several blockers of varying importance.
- Each user has personal preferences regarding working hours, importance of specific meetings, etc.
- QArrange is a Slack app, providing an app home tab and offering a chatbot.
- The chatbot can be tasked with the scheduling of calendar events with other participants.
- The chatbot memorizes personal calendar preferences of the user. These preferences can be viewed and managed on the Slack app home tab.
- QArrange has access to calendars and can create events after successful scheduling.
- A user queries the Secretary Agent via Slack, providing information about the meeting (title, desired date and time, duration, participants).
- The Secretary Agent makes sure that all necessary information is provided and asks for confirmation before proceeding. It then queries the Broker Agent.
- The Broker Agent requests available time slots from the individual User Agents and negotiates back and forth in case of conflicts, aiming for consensus.
- The User Agents (one per meeting participants) have access to user preferences and personal calendars and represent their users during the negotiation.
- After a working time slot has been reported back to the Secretary Agent, it will create a calendar event.
- A user tells the Secretary Agent about any kind of preference regarding their calendar.
- The Secretary Agent asks the Memory Agent to store the provided preference.
- The Memory Agent rephrases the preference and stores it in a database.
- Agno
- slack-sdk
- slack-bolt (in socket mode)
- Google Calendar
- SQLite (session storage, long-term memory)
- OpenAI language models
⚡️ Bolt app is running!
[U000000000A] App home opened.
[U000000000A-1763462485679699] Received message.
[U000000000A-1763462485679699-SecretaryAgent] Initializing.
[U000000000A-1763462485679699] Responded to message.
[U000000000A-1763462485679699] Received message.
[U000000000A-1763462485679699-SecretaryAgent] Initializing.
[U000000000A-1763462485679699-SecretaryAgent] Initialize scheduling of event:
Title: KI-Themen
Date: kommenden Donnerstag
Duration: 1 Stunde
Participants: ['U000000000A', 'U000000000B', 'U000000000C']
[U000000000A-1763462485679699-UserAgent-U000000000A] Initializing.
[U000000000A-1763462485679699-UserAgent-U000000000B] Initializing.
[U000000000A-1763462485679699-UserAgent-U000000000C] Initializing.
[U000000000A-1763462485679699-BrokerAgent] Initializing.
[U000000000A-1763462485679699-BrokerAgent] Negotiating.
[U000000000A-1763462485679699-UserAgent-U000000000A] Fetching calendar at 2025-11-20.
[GoogleCalendar] Fetching blocker events at 2025-11-20 from calendar a@group.calendar.google.com.
[GoogleCalendar] Fetched 2 blocker events.
[U000000000A-1763462485679699-UserAgent-U000000000B] Fetching calendar at 2025-11-20.
[GoogleCalendar] Fetching blocker events at 2025-11-20 from calendar b@group.calendar.google.com.
[GoogleCalendar] Fetched 1 blocker events.
[U000000000A-1763462485679699-UserAgent-U000000000C] Fetching calendar at 2025-11-20.
[GoogleCalendar] Fetching blocker events at 2025-11-20 from calendar c@group.calendar.google.com.
[GoogleCalendar] Fetched 1 blocker events.
[U000000000A-1763462485679699-SecretaryAgent] Creating event.
[GoogleCalendar] Creating event in calendar a@group.calendar.google.com: {'summary': 'KI-Themen', 'description': 'Created by QArrange.', 'start': {'dateTime': '2025-11-20T13:00:00+01:00'}, 'end': {'dateTime': '2025-11-20T14:00:00+01:00'}}
[GoogleCalendar] Created event: https://www.google.com/calendar/event?eid=...
[U000000000A-1763462485679699] Responded to message.
Make sure to have python3.10 available, e.g., via brew install python@3.10 on macOS. Using this version is a
workaround for an issue similar to this.
python3.10 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txtPrepare your environment in .env based on .env.example.
Furthermore, store credentials for a Google Cloud service account in google-calendar-service-account.json.secret. The account must be permitted to assume the role "Service Account Token Creator".
python main.pyRun the Agno Agent OS:
python agent_os.pyIn another terminal, run the Agno Agent UI:
npx create-agent-ui@latest
npm --prefix agent-ui run dev