Crew: handle meeting tasks

[whoabuddy]

Before each meeting I set up the notes and agenda:

• [ ] create a blank template based on past meetings
• [ ] fill in each section with relevant info
  • [ ] working group updates: any changes to the meeting notes, admin updates, crewAI updates
  • [x] latest AI news: summarizing content found over the week (so much, so fast!)
  • [ ] development updates: report progress on development activities, projects
  • [ ] demo time: plan for any live demos

With each meeting I copy over the following list of tasks to execute:

• [ ] download recording and transcript
• [ ] post meeting minutes on GitHub
• [ ] post video + recap on Twitter
• [ ] link recap in meeting minutes on GitHub
• [ ] post Twitter link to Stacks Discord
• [ ] post Twitter link to CrewAI Discord
• [ ] send email to RSVP list in AddEvent
• [ ] schedule tweets before next meeting

Right now I save the video and transcript locally, and it'd be great to explore a crew that can handle this step by step to help format things.

We can break down the meeting minutes into three sections:

• agenda items: one-line description and list of supporting information
• news items: one-line description with link and list of supporting information
• conversation:
  • capturing comments or discussions related to the items above
  • capturing open discussion and summarizing key points
  • properly and consistently crediting attendees / speakers

What we need to define in order to execute as a crew:

• what agent roles are needed to handle this? (a writer? a reviewer?)
• what tasks do we have for the agents?
• what do we want the final output to be?

It would be great if after a meeting we simply had to:

1. download the assets
2. kick off the crew
3. post the provided text to github/twitter

Then we can work on automating the posting part :japanese_ogre:

[whoabuddy]

Making some progress on this, new file prepare_for_meeting.py will search a list of links and summarize them, still some rough edges but a good step in the right direction. Also shows how to use the web scraping tool, pass context between tasks, and utilize gpt4o.

starting out and scraping a reddit page:

taking the raw content and turning it into something useful

finalizing it as a markdown list

Next up will be getting it to use the compiled final answer across all of the links, then generate the agenda item for latest AI news - it feels like it's getting closer!