Skip to main content
Threads are conversations with Humm. Each thread is a back-and-forth dialogue where you ask questions in plain English and Humm provides answers with data, charts, and recommendations.

​
Starting a Thread

Click New Thread from the sidebar to start a conversation. Type your question in the message box—no special syntax or query language needed. Example questions to get started:
  • “Which accounts have the highest churn risk?”
  • “Show me product usage trends for Acme Corp”
  • “What’s our average NPS score by customer segment?”

​
Understanding Humm’s Response

When you ask a question, Humm shows you two things:

​
Reasoning

The reasoning section shows how Humm is thinking through your question. You’ll see which data sources it’s querying, what calculations it’s running, and how it’s arriving at the answer. This transparency helps you trust the results and catch any misunderstandings early.

​
Answer

The answer section contains Humm’s response, which may include:
  • Text explanations — Context and insights about what the data shows
  • Charts and visualizations — Graphs that make patterns easy to spot
  • Data tables — The underlying numbers you can explore
  • Recommendations — Suggested next steps based on the analysis

​
Asking Follow-Up Questions

The real power of threads comes from follow-up questions. After Humm answers, you can dig deeper:
  • “Break this down by region”
  • “Why did churn spike in Q3?”
  • “Show me the accounts driving this number”
  • “Compare this to last quarter”
Humm remembers the context of your conversation, so you don’t need to repeat yourself.

​
Managing Threads

​
Finding Past Threads

All your threads are saved automatically. Use the sidebar to browse past conversations, or search by keyword to find a specific analysis.

​
Copying and Sharing

Every AI message has a Copy button to grab the text. Use the Share dropdown on a thread to share it with teammates — recipients can view the full conversation and continue asking their own questions.

​
Organizing Threads

Pin important threads to keep them at the top of your list. You can also rename threads to make them easier to find later.

​
Tips for Better Results

Instead of “Show me churn data,” try “Show me accounts that churned in the last 90 days with ARR over $50k.”
Humm defaults to reasonable timeframes, but specifying “last quarter” or “year over year” gets you exactly what you need.
Questions like “How does this compare to last month?” or “Which segment is performing best?” often reveal the most useful insights.
If you want a chart, table, or list, just ask. “Show me a trend chart of monthly active users” or “Give me a table of at-risk accounts.”

​
Uploading Files

Use the upload button in the chat input or drag a file into the conversation. Humm accepts CSV and XLSX files. Limits:
  • Up to 10 files per message
  • 10 MB per file
  • 50 MB total per message
After you upload, Humm shows a preview of the data before running any analysis. This is useful for ad-hoc datasets that aren’t in your warehouse — upload a spreadsheet and ask questions about it the same way you’d query a connected source.
Pair file uploads with a command to re-run the same analysis on fresh data each week. Export a report, drop it in, and let the command do the rest.

​
Attaching Images

Paste a screenshot or attach an image directly in the chat. Use this to share a chart you want Humm to recreate, point at a UI state, or paste an error message for troubleshooting.

​
Using @ References

Type @ in the chat input to reference specific tables, columns, or SaaS objects. Humm fetches their metadata (schema, row counts, sample values) and focuses the analysis on exactly what you’re pointing at. Syntax:
  • @[table:customers] — a database table
  • @[column:users.email] — a specific column
  • @[saas_object:Contact] — a CRM/SaaS object type
  • @[saas_field:Contact.owner] — a specific field on an object
This is especially useful when your schema has similarly named tables or when you want Humm to skip discovery and go straight to the right data.

​
When to Start a New Thread

Start a new thread when you’re switching to a completely different topic. If you’re still exploring the same general area, keep going in the same thread. The context helps Humm give better answers.