HR Policy Q&A Bot: Telegram + Groq + Gemini + Drive (RAG)

Ask your HR questions—automatically—directly in Telegram (with RAG over your HR policy)

This n8n automation workflow indexes your HR policy PDF from Google Drive and answers employee questions in Telegram using a Groq chat model with retrieval-augmented generation (RAG). No manual searching—just grounded answers pulled from your policy text.

What this workflow does

• Loads your HR policy PDF: Manually downloads the HR policy document from Google Drive.
• Builds a searchable knowledge base: Splits the PDF content, generates Google Gemini embeddings, and inserts chunks into an in-memory vector store.
• Listens in Telegram: Triggers whenever a new message is received by your Telegram bot.
• Retrieves relevant policy passages: Uses the vector store to find relevant excerpts for the user’s question.
• Generates a grounded response: Calls a Groq chat model with per-chat conversation memory, then sends the answer back to the originating Telegram chat.

Use cases

• Employees ask HR questions like leave policy, reimbursements, or code-of-conduct rules, and receive answers sourced from the policy.
• SaaS operators reduce HR workload by handling repetitive “policy lookup” questions via Telegram.
• Automation engineers create a maintainable HR Q&A assistant that re-indexes when the PDF changes.

Technical details

• Telegram: Telegram bot token (via @BotFather) + Telegram trigger for incoming messages.
• Google Drive: OAuth credentials to download the HR policy PDF.
• Google Gemini: API key used for embeddings during indexing.
• Groq: API key powering the chat model for RAG responses.
• n8n / LangChain Agent: Uses an “HR Policy Agent” node to retrieve passages and generate answers; you can edit its system message to adjust tone, language, or word limit.
• Important note: Run the manual “Load knowledge base” trigger once to index, and re-run it when the document changes or after n8n restarts (the in-memory store resets).