Find Reddit threads on reader pain points
Testing if Comet can support real newsletter work with copy-paste prompts across 4 workflows. Task: The author used Perplexity Comet to automate data ingestion for NotebookLM by searching Reddit for user pain points related to traditional browsers. Comet executed a prompt to find relevant threads, manually input URLs, and utilize NotebookLM's 'Discover' feature to add sources. The agent operated autonomously for approximately 15 minutes without human intervention during the search and addition process.; tools: Perplexity Comet. Human role: The original post does not say how a person stayed involved.; limitations: This workflow is recommended only when the user is not committed to specific sources and requires broad web discovery. It may be less efficient than traditional browsing if the user already has a curated list of links. The source does not specify error handling for broken links or irrelevant results during the autonomous execution..
Tools
Published on Agent Work Atlas:
Overview
Testing if Comet can support real newsletter work with copy-paste prompts across 4 workflows.
Workflow
The author used Perplexity Comet to automate data ingestion for NotebookLM by searching Reddit for user pain points related to traditional browsers. Comet executed a prompt to find relevant threads, manually input URLs, and utilize NotebookLM's 'Discover' feature to add sources. The agent operated autonomously for approximately 15 minutes without human intervention during the search and addition process.
Reported result
Comet successfully identified more Reddit threads than requested and added 10 additional sources via the 'Discover' feature. The task completed in roughly 15 minutes with no intervention required. The author noted this workflow is best when specific sources are not pre-curated and extensive web search is needed.
Human role
The original post does not say how a person stayed involved.
Limitations
This workflow is recommended only when the user is not committed to specific sources and requires broad web discovery. It may be less efficient than traditional browsing if the user already has a curated list of links. The source does not specify error handling for broken links or irrelevant results during the autonomous execution.
Try it
Act as a research assistant using Perplexity Comet. My goal is to populate a NotebookLM notebook with sources discussing [INSERT TOPIC, e.g., AI productivity complaints]. 1. Search Reddit for [INSERT NUMBER, e.g., 20] threads discussing [INSERT PAIN POINTS]. 2. Use the 'Add Source' feature in NotebookLM to input these URLs. 3. Trigger NotebookLM's 'Discover' feature to find additional related sources. 4. Add all discovered sources to the notebook. Before executing, confirm the topic and source count with me. Run a dry test with 3 sources first to verify the workflow. If successful, proceed with the full batch. Report the total number of sources added and any errors encountered. Do not publish or send any data externally; keep all actions within the Comet and NotebookLM interface.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.