> ## Documentation Index
> Fetch the complete documentation index at: https://docs.humyn.studio/llms.txt
> Use this file to discover all available pages before exploring further.

# Add Comments

> Seed Humyn with the sources, keywords, and topics that drive each insight run

# Add Comments

<img src="https://mintcdn.com/humyn/SR6kd4QqpyH32m4a/screenshots/add_comments.png?fit=max&auto=format&n=SR6kd4QqpyH32m4a&q=85&s=28026e77d717d7394a43adab776e3bef" alt="Humyn Add Comments interface" style={{ borderRadius: "1rem", margin: "2rem 0", boxShadow: "0 24px 60px rgba(15, 23, 42, 0.16)" }} width="1563" height="990" data-path="screenshots/add_comments.png" />

## Define your research topic

Every run starts with a clear <mark>research topic</mark>. Use a short sentence that tells the Humyn ML engine exactly what to prioritize—think “onboarding friction for AI teams” or “why creators migrate off Platform X.” This keeps clustering and sentiment analysis pointed at the signal you care about.

## Seed strategic keywords

List the <mark>keywords</mark> and phrases Humyn should listen for. Mix explicit product terms with intent-led language (“upgrade flow,” “pricing jump,” “workflow automation”). You can add or edit the list anytime before you launch a run.

## Drop source URLs

Paste one or many <mark>Reddit URLs</mark> (plus other supported sources) into the feed. Humyn will fetch full threads, comments, and context automatically. Batch pasting works great when you have a backlog from research sprints or community callouts.

## Run the analysis

Hit <mark>Run analysis</mark> to kick off ingestion. Humyn spins up a processing run, shows real-time progress, and marks the job as <mark>100% processed</mark> once everything is clustered and tagged. From there you can jump straight into <mark>Explore</mark>, <mark>Metrics</mark>, or <mark>Chat with Hue</mark> to work with the new data.