Once a day the scenario pulls tickets from Zendesk, loads them into the Postgres PGVector Store search base, finds the recurring themes and posts a digest to Twake.
No need to read every ticket
You see recurring complaints, not isolated tickets
The digest lands in Twake daily
Ticket history builds up in the search base
Postgres PGVector StoreTwakeZendeskPinecone Vector StoreEmbeddings OpenAIDefault Data Loader