Hi! Iβm trying to do a specific task with AI. Being a marketing girl, Iβm asking for advise for the best set up to build it for the engineers in the room π The usecase:
On one side, I have a list of 1000 persons of a specific ecosystem.
One the other side, I have a list of 40-50 topics.
I want to use AI to help me spot the best person to talk about the specific topic. The purpose is then to invite each person on a podcast on each specific topic.
Today, I have an assistant design to help me prepare an episode. Heβs doing a great job when I tell him the name of one guest. But I tried to give him a list of 10 names and ask for topics for theses person and the answer wasnβt really good. So Iβm wondering, which set up should I do to improve the answer? Thanks for your help β¨
Hello Tifany Clemenceau π Here is what I would personally do :
Create a "Person overview" assistant - the assistant would be task to give you an overview of someone's work & expertise (can have web browsing)
List all potential speakers in a Gsheet
Use Dust in Gsheet to have the people's description filled by Dust in Column B (using the column A as input)
Create another assistant that uses the name & overview as input to assign top 3-4 topics based on your list, to add them in column C with "," separating each topic. You might need to add things like "ONLY give the topcis in your answer. Never give an intro".
You can then export your Gsheet and re-import it in Notion which should give you a clean & searchable table with topics showing as "tags" that you can filter/search through
Docs:
Dust in Gsheet: https://docs.dust.tt/docs/beta-google-spreadsheets
Notion import csv: https://www.notion.com/help/import-data-into-notion#csv
Let me know if you got any question ! If others have alternative approaches, feel free to jump in π
