Weekly AI Trend Alert System using Research Papers and Vector Search - n8n Workflow

• AI/ML researchers and developers needing concise weekly trend summaries.


• Data scientists and engineers looking for practical n8n templates involving vector databases.


• Users wanting to implement complex agentic Retrieval-Augmented Generation (RAG) within an n8n node structure.


• Anyone looking to build custom, scheduled data enrichment and reporting pipelines using n8n.

The challenge of staying updated with cutting-edge AI research published on platforms like arXiv is immense. This advanced n8n workflow provides a solution by creating a fully automated weekly trend alerter. This specific n8n workflow starts with a scheduled n8n trigger, pulls the latest papers, uses a large language model (LLM) for semantic enrichment (classifying topics and predicting impact), and then stores this enriched data efficiently in a Weaviate vector database. The true innovation lies in the subsequent step, where an n8n agentic RAG system queries the vector store to analyze the collective dataset, identify overarching trends, cite relevant papers, and compile a polished, summarized email report. By leveraging powerful n8n node functions for data transformation, LLM interaction, and vector search, this setup moves far beyond simple scraping, offering deep, contextual analysis. This n8n template is essential for anyone serious about tracking the pulse of the AI industry.

This comprehensive n8n workflow executes in two main phases, triggered weekly by the Schedule n8n node:

1. Data Acquisition and Vectorization:

1. Import the n8n Workflow: Copy the provided JSON code and import it directly into your self-hosted n8n instance via the "New" button in the workflow editor.


2. Weaviate Credentials: You must set up a Weaviate credential pointing to your cloud cluster or local instance. This is required for both the insertion (Weaviate Vector Store) and retrieval (Weaviate Vector Store1) n8n nodes. Ensure the collection name ArxivArticles is consistent across the n8n workflow.


3. LLM Credentials (OpenAI/OpenRouter): This n8n workflow uses both OpenAI and OpenRouter credentials for different LLM operations (embeddings and chat models). Configure these API keys within your n8n environment.


4. SMTP Setup: Set up an SMTP credential (e.g., Gmail) to enable the final Send email n8n node. Specify the sender and recipient email addresses in the Send email node parameters.


5. Activation: Once all credentials are set up in the respective n8n node connections, ensure the n8n workflow is active and awaiting the weekly schedule n8n trigger.

Schedule Trigger n8n trigger: The starting n8n node, configured to run the entire n8n workflow once per week.
Query arXiv (HTTP Request n8n node): Fetches up to 200 recent papers categorized as cs.LG or stat.ML, dynamically setting the date range based on the preceding Date & Time n8n node output.
Prep Data for Weaviate (Set n8n node): Transforms complex XML fields (like nested authors and categories) into clean JSON arrays and ISO datetime formats, optimizing data structure before enrichment.
Enrich Articles with Topic Classification (AI Agent n8n node): A powerful n8n node that uses an LLM (via OpenRouter/OpenAI) and a Structured Output Parser to categorize papers into defined topics and assign a potential_impact score (1-5).
Weaviate Vector Store (n8n node): Handles the core function of indexing and inserting the enriched documents into the Weaviate collection (ArxivArticles), generating embeddings for the summary field using the Embeddings OpenAI n8n node.
Agentic RAG for Trend Analysis (AI Agent n8n node): The primary analytical n8n node. It is configured with the Weaviate Vector Store1 tool for RAG. Its comprehensive system message dictates trend identification criteria, focusing on topic counts and high-impact papers, and strictly enforces the final email JSON output structure. This sophisticated use of an n8n node facilitates deep automated analysis.
Structured Output Parser (n8n node): Used twice, this essential n8n node guarantees that the LLM output conforms exactly to the required JSON schema, defining the fields for enrichment (Part 1) and the final email format (Part 2).
Send email (n8n node): The final action, using SMTP credentials to send the HTML-formatted weekly report generated by the preceding Markdown n8n node, concluding the n8n workflow.