Ask about your data, then run the analysis.
A walkthrough on a synthetic dataset, recorded with the live AI model. Captions are on by default, and the full transcript is under the video.
Transcript
Celeus: ask a question about your data, then run the analysis it proposes.
We upload a small synthetic dataset: forty rows, a treatment arm and an outcome.
We open the chat. The model is sent a description of the dataset, such as column names, summary statistics and category names, with small counts suppressed, plus the dataset name, the results of analyses, and the text we write, including chat messages. It is not sent the uploaded file or the data rows.
We ask in plain language whether the outcome differs between the arms, and for an analysis we can run.
The reply comes with a proposed analysis card, naming the analysis it suggests. Nothing has been added to our results yet: the choice to run it is ours.
We choose Run now. The analysis starts in R without leaving the page, and the notification opens the run. The wait is shortened in this video.
Here is the result, computed in R by the same analysis flow as a run set up by hand. As with every run, the page asks us to verify the result before we accept it.
Where to go next
- Video: from upload to package
Upload a dataset, run a suggested method in R and download the reproducible package.
- Getting started
Upload a dataset and run your first analysis.
- Guides
How to choose a statistical method for your question.
- Security and data handling
What reaches the AI, and what never does.
- Pricing
Every plan gets every method.