Zum Inhalt springen
NEUMarktplatz und dein Team: Verkaufe, was du baust, und bekomm Pläne und Änderungen, die getestet sind, bevor du sie siehst. Neuigkeiten ansehen

Bibliothek / RAG & Wissensdatenbanken

RAG & Wissensdatenbanken

HR-Richtlinienfragen in Telegram beantworten mit Groq, Gemini und Google Drive

Manuell ausführen, um eine HR-Richtlinien-PDF aus Google Drive in einen Vektorspeicher aufzunehmen; danach erhält jede Telegram-Nachricht eine Antwort von einem KI-Agenten, die auf dem Richtlinientext basiert.

Was sie bewirkt

## Answer HR policy questions in Telegram with RAG (Groq, Gemini embeddings, Google Drive) ### How it works This workflow builds a retrieval-augmented HR policy assistant for Telegram. A manual ingestion path downloads an HR policy PDF from Google Drive, parses it, creates Gemini embeddings, and inserts the content into an in-memory vector store. When a Telegram message arrives, an AI agent uses Groq, chat memory, and the policy vector store to answer the employee’s question, then sends the response back to Telegram. ### Setup steps - Configure Google Drive credentials and point the download node to the HR policy PDF file. - Configure Google Gemini credentials for both embedding nodes used during insert and retrieval. - Configure Groq credentials for the chat model used by the HR Policy Agent. - Configure Telegram bot credentials for the trigger and reply nodes, including the correct chat/message fields. - Run the manual knowledge-base loading path before answering questions so the in-memory vector store is populated. ### Customization Swap the Google Drive file for another policy document, adjust the agent prompt and Groq model, tune PDF loading/chunking behavior, or replace the in-memory vector store with a persistent vector database if the index must survive workflow restarts.

Die Schritte

  1. When Manually Triggered
  2. Download PDF from Google Drive
  3. Store in Vector Database
  4. Generate Gemini Embeddings
  5. Load PDF Data
  6. When Message on Telegram
  7. HR Q&A Agent
  8. Send Reply on Telegram
  9. Groq Chatbot Model
  10. Memory Buffer for Chat
  11. HR Policy Vector Storage
  12. Retrieve Gemini Embeddings

Ähnliche Automatisierungen