Give your team a smart helper inside Slack. Teammates mention the bot in a channel and get fast answers from your company files. It works well for IT, HR, and operations FAQs where people ask the same questions every day.

The flow listens to app mentions in Slack and passes the message to an AI agent powered by Anthropic. A short chat memory is kept per channel and user, so follow up questions stay in context. A retrieval tool searches a Qdrant vector database for the top matching document chunks, then the agent writes the final reply. Your Google Drive files are loaded, split into small parts, embedded with OpenAI, and saved into Qdrant. HTTP calls create and refresh the Qdrant collection, and replies return to Slack in a thread with clean formatting.

Setup needs a Slack app with specific scopes, OpenAI and Anthropic API keys, a Qdrant URL and key, and Google Drive access. After indexing a chosen folder, the bot answers common requests in seconds and lowers ticket volume. Use it for onboarding, policy lookups, step by step procedures, and simple troubleshooting across teams.

• Slack app mention trigger captures questions and posts answers back in the same thread.
• AI agent uses Anthropic Chat Model to write clear, helpful replies.
• Short term memory per channel and user keeps the last 10 turns for better follow ups.
• RAG tool queries Qdrant and returns the top 10 matching document chunks.
• Google Drive ingestion loads a folder, downloads files, and prepares them for search.
• Token splitter breaks documents into small parts for better retrieval.
• OpenAI embeddings convert text into vectors for both indexing and searching.
• HTTP steps create and refresh the Qdrant collection before loading new content.

• Reduce manual question handling from 1 hour to 1 minute per request
• Automate up to 70% of common internal questions
• Cut response time from minutes to seconds with instant answers in Slack
• Improve answer consistency by using one up to date document source
• Connect Slack, Google Drive, Qdrant, OpenAI and Anthropic in one flow
• Support many users at once with threaded replies in channels

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Slack, Google Drive, Qdrant, OpenAI and Anthropic. See the Tools Required section above for links to create accounts with these services.
3. In your Slack workspace, create a Slack app. Add event subscriptions for app_mention and message.channels. Grant scopes including app_mentions:read, channels:history, channels:read, chat:write, files:read, groups:history, im:history, mpim:history, reactions:read, reactions:write, users:read. Install the app to the target channel.
4. In n8n, open the Slack Trigger node and the Slack Send node. For each, choose Credential to connect with, click Create new credential, and follow the on screen steps to connect your Slack app token. Select the target channel in both nodes.
5. Set up Qdrant. Get your Qdrant URL and API key. In n8n, open the HTTP Request nodes Create collection and Refresh collection and replace QDRANTURL and COLLECTION with your values. For authentication, create an HTTP Header credential and include the API key if your Qdrant setup requires it.
6. Open the Qdrant Vector Store nodes and the RAG node. Set the Collection name to match your Qdrant collection. Confirm the Qdrant credential is selected.
7. Connect OpenAI. Open both Embeddings nodes, pick Credential to connect with, click Create new credential, and paste your OpenAI API key from the OpenAI API page.
8. Connect Anthropic. Open the Anthropic Chat Model node, choose Credential to connect with, click Create new credential, and paste your Anthropic API key from the Anthropic account page.
9. Authorize Google Drive. Open the Google Drive Get folder and Download Files nodes. Create a new Google Drive OAuth2 credential and allow access. Choose the folder that holds your company documents.
10. Check the data loader path. Confirm Default Data Loader and Token Splitter are connected before the Qdrant Vector Store node so files are parsed and chunked correctly.
11. Build the index. Click Execute on the Manual Trigger path to run Create collection, Refresh collection, download files, and write vectors to Qdrant. Verify successful upserts in the execution log or your Qdrant dashboard.
12. Test in Slack. In the target channel, mention the bot with a question the documents can answer. Confirm the reply appears in a thread and uses information from your files.
13. Troubleshoot common issues. If no replies, check Slack scopes and channel ID. If search fails, confirm QDRANTURL and collection name. If indexing stops, review Google Drive permissions and file types. For poor answers, increase or decrease the token chunk size or adjust topK in the RAG node.