This content originally appeared on DEV Community and was authored by VICTOR LAKRA
This is a submission for the AI Agents Challenge powered by n8n and Bright Data
What I Built
I built a JARVIS-inspired AI Agent workflow in n8n that acts as a smart assistant capable of handling two core tasks:
- Fetching the latest news using a dedicated
news_agentworkflow. - Analyzing Crunchbase startups/companies based on keyword-driven queries via the
analyze_startupsworkflow.
It also includes a fun wake-up activation feature: when the user says “Wake up, daddy’s home.” it triggers music playback from Spotify, just like summoning JARVIS in Iron Man style. This combines AI-powered intelligence with practical and entertaining automation.
Demo
Here’s a walkthrough of the JARVIS AI Agent in action:
- You can issue a query like “Analyze fintech companies on Crunchbase” and the workflow extracts the keyword and calls the
analyze_startupssub-workflow. - A query like “Get me the latest news” routes to the
news_agentworkflow, which scrapes BBC, enriches with Bright Data, and formats results into a witty newsletter-style digest. - Saying “Wake up, daddy’s home.” makes it greet you by starting a Spotify playlist.
n8n Workflow
You can view the full workflow JSONs here:
Technical Implementation
The agent is built using n8n’s AI and automation nodes:
- System Instructions: A strict prompt ensures JARVIS only handles news and Crunchbase analysis, rejecting unrelated queries.
-
Model Choice:
gpt-4o-minipowers the conversational reasoning. - Memory: A buffer memory node maintains conversational context per session.
-
Tools: Two tool workflows —
news_agentfor news,analyze_startupsfor startup analysis. - Webhook: Handles user input and responds back with AI-driven results.
- Spotify Node: Adds the interactive wake-up feature for user delight.
News Agent
- Pulls BBC headlines via RSS.
- Uses Bright Data to scrape and enrich article content.
- Aggregates and formats results.
- Sends to an AI Agent node with instructions to rewrite in a Morning Brew–style witty newsletter.
- Delivers the final email digest via Gmail integration.
Analyze Startups Agent
- Triggers a Bright Data Crunchbase dataset search based on a keyword (e.g., fintech, AI, climate tech).
- Polls until the snapshot is ready.
- Retrieves startup/company details.
- Cleans and sorts data with Python code (top 10 by most recent founded date).
- Returns a concise, structured summary with insights on funding, team, products, and market signals.
Bright Data Verified Node
Both news_agent and analyze_startups integrate Bright Data’s verified data nodes for reliable scraping and enrichment. This ensures that the insights pulled from external data sources are accurate, fresh, and robust.
Why Real-Time Data Mattered
Real-time data made JARVIS genuinely useful instead of just being a static demo. Fresh headlines gave the news digest actual relevance for the day, and pulling Crunchbase startups dynamically ensured users could explore current market players rather than outdated datasets. Without real-time enrichment, the workflows would have felt like canned responses instead of a living, adaptive assistant.
Journey
The journey started with wanting to create a JARVIS-like AI that’s both useful and fun. I faced challenges in:
- Designing a clear system prompt so the AI never strays outside its defined scope.
- Ensuring keyword extraction for Crunchbase queries worked consistently.
- Managing session memory to give the agent contextual awareness.
- Formatting news into a polished HTML digest that feels engaging.
Key learnings:
- n8n’s agent node makes it seamless to route user queries to the right tools.
- Bright Data’s verified node is essential for scaling reliable web data into AI pipelines.
- Adding small touches like Spotify integration creates a delightful user experience.
This project brought together AI, automation, and a touch of sci-fi imagination into one cohesive workflow.
This content originally appeared on DEV Community and was authored by VICTOR LAKRA