Smart RSS Feed Filtering and Baserow Tracking - n8n Workflow

Content Curators: Users who need to monitor high-volume RSS feeds without duplicating alerts.
Technical Analysts: Professionals tracking industry news or specific competitor updates.
n8n Developers: Anyone looking for comprehensive n8n templates demonstrating advanced AI agent integration and database state management.
Small Businesses: Teams needing reliable, custom news aggregation delivered directly to Slack.

Information overload is a major challenge when tracking multiple RSS feeds. Standard deduplication relies on simple GUID checks, but ensuring state persistence across runs requires robust database integration. This advanced n8n workflow solves this by creating a highly resilient monitoring system.

The core value of this n8n automation lies in its integration of Baserow for state management and an OpenAI-powered AI Agent for intelligent filtering. Every time the n8n trigger runs, the system compares incoming articles against a comprehensive list of previously processed items stored in Baserow. The AI Agent then cleanly outputs only the truly new articles, minimizing noise and ensuring your Slack channel receives only actionable updates. This sophisticated approach makes this one of the most powerful n8n templates for content monitoring.

This n8n workflow operates in several distinct stages, ensuring highly accurate filtering and tracking:

1. Workflow Initiation: The process starts with a manual n8n trigger (or a scheduled one if configured).


2. Configuration Retrieval: The first Baserow n8n node, 'Read Rss Link', fetches the list of target RSS feed URLs from a designated configuration table in Baserow. The items are then split out using the 'Split Out' n8n node.


3. Data Fetching & Parsing: For each URL, the 'Fetch HTML' node retrieves the XML data, which is immediately converted into a usable JSON format by the 'XML Converter' n8n node.


4. State Retrieval: Concurrently, the 'Get Seen Products' Baserow n8n node retrieves all previously recorded article links (GUIDs) from the 'Seen Products' table.


5. Data Preparation: The 'Edit data structure' n8n node combines the newly fetched articles and the list of 'seen' GUIDs into a single, structured payload for the AI model.


6. AI Filtering: The powerful 'AI Agent' n8n node, leveraging the 'OpenAI Chat Model' (GPT-4o Mini) and Langchain memory, analyzes the combined data. The prompt specifically instructs the AI to return only the articles whose GUID is not present in the alreadyProcessedGuids array, acting as a smart deduplicator.


7. Output Processing: The 'Clean JSON' n8n node parses the raw JSON string output provided by the AI Agent, transforming the resulting new articles back into individual n8n items.


8. State Update: For every newly processed article, the 'Save seen products' Baserow n8n node creates a record in the 'Seen Products' table, updating the workflow's state and preventing future duplicates.


9. Notification: Finally, the 'Slack' n8n node sends a well-formatted notification containing the article's title, content snippet, and link to the configured Slack channel.

To deploy this comprehensive n8n workflow, follow these steps:

1. Import the n8n template: Copy the provided JSON data and paste it into your n8n instance via the 'Import from JSON' option.


2. Baserow Credentials: Configure your Baserow API Key credential. You must update the Database and Table IDs in the 'Read Rss Link', 'Get Seen Products', and 'Save seen products' n8n nodes to match your Baserow setup, ensuring the correct rssLink and Nom fields are used for input and output, respectively.


3. OpenAI Credentials: Set up your OpenAI API key credential. This is required for the 'OpenAI Chat Model' n8n node that powers the AI Agent filtering.


4. Slack Credentials: Configure your Slack API or Webhook credentials and ensure the 'Slack' n8n node is targeting the correct channelId (e.g., C091X7ZNW4V in the provided configuration).


5. Review the AI Prompt: Inspect the 'AI Agent' n8n node. While the prompt is designed for this task, you can customize it further if your article data structure differs.


6. Activation: Set the n8n workflow to active. Replace the 'Click to Start' n8n trigger with a time-based 'Schedule Trigger' for automated monitoring.

This n8n workflow relies on several specialized n8n node types to achieve its goal:

Click to Start (n8n trigger): The starting point for execution, typically replaced by a scheduled n8n trigger for production monitoring.
Read Rss Link & Get Seen Products (Baserow n8n node): These nodes manage the workflow's state and configuration, reading the list of feeds to monitor and the GUIDs of articles already processed.
Fetch HTML & XML Converter (HTTP Request / XML n8n node): The workflow uses these nodes sequentially to download the raw RSS feed data and convert the complex XML structure into a manageable JSON format.
Edit data structure (Code n8n node): Performs crucial data shaping, packaging the new articles and the array of alreadyProcessedGuids into the specific format required by the downstream AI Agent.
OpenAI Chat Model (Langchain n8n node): Provides the large language model (GPT-4o-mini) capability necessary for the AI Agent to perform logical comparison and filtering.
AI Agent (Langchain n8n node): The key filtration logic. This powerful n8n node executes instructions against the input data, returning only the articles that are genuinely new, based on the provided list of seen GUIDs.
Clean JSON (Code n8n node): Parses the raw JSON string output returned by the AI Agent, preparing the newly filtered items for subsequent database storage and notification steps. This is a vital step when integrating external language model outputs into the main n8n flow.
Save seen products (Baserow n8n node): Creates a record in the database for each successfully processed new article, maintaining the integrity of the tracking system.

• Slack (Slack n8n node): The final destination for the filtered content, sending structured alerts to the designated channel.