Skip to main content
Follow these steps to take one question from your data to a writeup whose every claim traces back to evidence. Along the way you get a notebook of the analysis, parallel ideas the agent tests, and knowledge that later work builds on. Each step builds on the previous one.
1

Download Qualia

Get the desktop app from quadrillion.ai/download. For per-platform install steps, see Getting Started. To skip the install, open Qualia Cloud instead.
2

Open your workspace

Launch Qualia and open the folder where your data lives. A workspace is the folder Qualia works in, and the agent can read every file in it. For example, CSV, Parquet, and Excel files in that folder are available right away, with no import step. If the data sits on another machine, connect to a remote backend.
3

Start with a prompt

Describe what you want in the chat box, for example “explore sales_q3.csv”. If the file isn’t in your workspace yet, attach it with the paperclip. To start from an example instead, choose one of the example prompts below the chat box.
Empty Qualia workspace with the chat box and example prompt cards
4

Watch the analysis in a notebook

The agent loads your data and runs the analysis in a Jupyter-compatible notebook. It narrates what it finds in the chat panel as it goes. Every cell is ordinary, editable code, so you can rerun it, change it, or add your own. You can also describe a change in chat and let the agent make it.
Agent writing and running notebook cells with a summary in the chat panel
5

Follow parallel ideas

For bigger questions, Qualia splits the work into ideas. An idea tests one hypothesis or approach, and Qualia gives each idea its own agent. The ideas appear in the Chats list in the left sidebar, beneath the chat that proposed them. Choose one to follow its agent’s work. See Ideas and tasks.
To change how readily Qualia splits work across agents, choose Research style in the chat box and adjust Parallelism. See Parallelism.
6

Run a skill

A skill is a saved workflow you invoke on demand, so you don’t re-explain a process each time. Type / in the chat box to see the list. Start with a built-in skill such as /dataset-exploration or /optimize.
Skill picker opened with the slash key in the chat box
You can also write your own skills, or import the ones you have from Claude Code or Cursor. To browse skills, rules, MCP servers, and more, open the gear menu and choose Customize agent.
Skills library open in the Customize agent tab
7

Get a research writeup

Ask in chat for a write-up of the findings. You get a writeup, a research document with findings, figures, and numbered citations. Edit it alongside the notebook, and export it when it’s ready to share.
Generated writeup with figures and numbered citations
8

Trace any claim to its evidence

Choose a citation in the writeup to preview its supporting evidence. Then choose Open in Knowledge to inspect the claim in Knowledge. A claim is a finding Qualia records with links to the evidence behind it. Knowledge persists across sessions, so later work builds on what earlier work established instead of rediscovering it.

Next steps

  • Chat & AI: prompting, models, and chat features
  • Notebooks: cell operations, keyboard shortcuts, and exports
  • Ideas and tasks: how agents split and track work
  • Knowledge: claims, evidence, and recall priority
  • Agents: running parallel experiments
  • Skills: save, write, and import reusable workflows
  • Writeups: research documents with citations
  • Autonomous mode: longer unattended runs with Slack updates
  • Rules: persistent instructions for the agent
  • Settings: models, appearance, and more