Baserow and Google Gemini Chat Model integration

This n8n workflow template simplifies the process of digesting cybersecurity reports by summarizing, deduplicating, organizing, and identifying viral topics of interest into daily emails. It will generate two types of emails: A daily digest with summaries of deduplicated cybersecurity reports organized into various topics. A daily viral topic report with summaries of recurring topics that have been identified over the last seven days. This workflow supports threat intelligence analysts digest the high number of cybersecurity reports they must analyse daily by decreasing the noise and tracking topics of importance with additional care, while providing customizability with regards to sources and output format. How it works The workflow follows the threat intelligence lifecycle as labelled by the coloured notes. Every morning, collect news articles from a set of RSS feeds. Merge the feeds output and prepare them for LLM consumption. Task an LLM with writing an intelligence briefing that summarizes, deduplicates, and organizes the topics. Generate and send an email with the daily digest. Collect the daily digests of the last seven days and prepare them for LLM consumption. Task an LLM with writing a report that covers 'viral' topics that have appeared prominently in the news. Store this report and send out over email. How to use & customization The workflow will trigger daily at 7am. The workflow can be reused for other types of news as well. The RSS feeds can be swapped out and the AI prompts can easily be altered. The parameters used for the viral topic identification process can easily be changed (number of previous days considered, requirements for a topic to be 'viral'). Requirements The workflow leverages Gemini (free tier) for email content generation and Baserow for storing generated reports. The viral topic identification relies on the Gemini Pro model because of a higher data quantity and more complex task. An SMTP email account must be provided to send the emails with. This can be through Gmail.

This workflow finds fresh Reddit posts that match your keywords, decides if they’re actually relevant to your brand, writes a short human-style reply using AI, posts it, and logs everything to Baserow. 💡Perfect for Lead gen without spam: drop helpful replies where your audience hangs out. Get discovered by AI surfaces (AI Overviews / SGE, AISEO/GSEO) via high-quality brand mentions. Customer support in the wild: answer troubleshooting threads fast. Community presence: steady, non-salesy contributions in niche subreddits. 🧠 What it does Searches Reddit for your keyword query on a schedule (e.g., every 30 min) Checks Baserow first so you don’t reply twice to the same post Uses an AI prompt tuned for short, no-fluff, subreddit-friendly comments Softly mentions your brand only when it’s clearly relevant Posts the comment via Reddit’s API Saves post_id, comment_id, reply, permalink, status to Baserow Processes posts one-by-one with an optional short wait to be nice to Reddit ⚡ Requirements Reddit developer API Baserow account, table, and API token AI provider API (OpenAI / Anthropic / Gemini) ⚙️ Setup Instructions Create Baserow table Fields (user-field names exactly): post_id (unique), permalink, subreddit, title, created_utc, reply (long text), replied (boolean), created_on (datetime). Add credentials in n8n Reddit OAuth2* (scopes: read, submit, identity) and set a proper User-Agent* string (Reddit requires it). LLM**: Google Gemini and/or Anthropic (both can be added; one can be fallback in the AI Agent). Baserow**: API token. Set the Schedule Trigger (Cron) Start hourly (or every 2–3h). Pacing is mainly enforced by the Wait nodes. Update “Check duplicate row” (HTTP Request) URL**: https://api.baserow.io/api/database/rows/table/{TABLE_ID}/?user_field_names=true&filter__post_id__equal={{$json.post_id}} Header**: Authorization: Token YOUR_BASEROW_TOKEN (Use your own Baserow domain if self-hosted.) Configure “Filter Replied Posts” Ensure it skips items where your Baserow record shows replied === true (so you don’t comment twice). Configure “Fetch Posts from Reddit” Set your keyword/search query (and time/sort). Keep User-Agent header present. Configure “Write Reddit Comment (AI)” Update your brand name** (and optional link). Edit the prompt/tone** to your voice; ensure it outputs a short reply field (≤80 words, helpful, non-salesy). Configure “Post Reddit Comment” (HTTP Request) Endpoint: POST https://oauth.reddit.com/api/comment Body: thing_id: "t3_{{$json.post_id}}", text: "{{$json.reply}}" Uses your Reddit OAuth credential and User-Agent header. Update user_agent value in header by your username n8n:reddit-autoreply:1.0 (by /u/{reddit-username}) Store Comment Data on Baserow (HTTP Request) POST https://api.baserow.io/api/database/rows/table/{TABLE_ID}/?user_field_names=true Header: Authorization: Token YOUR_BASEROW_TOKEN Map: post_id, permalink, subreddit, title, created_utc, reply, replied, created_on={{$now}}. Keep default pacing Leave Wait 5m (cool-off) and Wait 6h (global pace) → \~4 comments/day. Reduce waits gradually as account health allows. Test & enable Run once manually, verify a Baserow row and one test comment, then enable the schedule. 🤝 Need a hand? I’m happy to help you get this running smoothly—or tailor it to your brand. Reach out to me via email: [email protected]