Overview
Glean is launching Data Analysis for Structured Files in Assistant.
This feature will allow users to analyze files, starting with CSV, XLSX and JSON uploads. Users across finance, sales and support, IT can save hours on ad hoc analyses.
Some questions you can ask:
Dataset Exploration, i.e. “What columns does my data contain?”
Column Statistics, i.e. “average”, “sum”, “count”
Data Distribution, i.e. “Top / bottom x values”, “What is the range of values in x column”, “Show me all the unique values in x column”
Data Relationships, i.e. “How does conversion change by lead source?”
Time-Series Relationships, i.e. “What product area have we gotten more tickets in over the last month compared to previous months”
Key Features
Assistant UX: You can do Data Analysis on either uploaded files or indexed files by tagging them or pasting in their link in Assistant. You can also generate files as outputs. You can ask multiple questions back and forth in multi-turn fashion.
API Support: Customers who utilize our developer platform can review the File Upload APIs documentation or visit developers.glean.com
Security: Data Analysis for Structured Files is secure at all stages of the file lifecycle, since analysis is done in a sandboxed way for each user. For more, please see the security whitepaper
Supported File Formats
Spreadsheet Files: xls, xlsx
Text Files: csv, json
Known Limitations
Multi-media support: We do not support generating visualizations in this release.
Number of files: We support single-file uploads only. A single XLS/XLSX file can contain multiple tabs.
Custom data retention policies: Per the File Upload feature, we do not support data retention policies beyond the default 24 hour policy and 30 days policy for metadata described above. You can ask users to disable chat session history or to manually delete chat sessions if you would like to delete metadata sooner.
Complex sheets: Understanding complex sheets containing merged cells and hierarchical tables, charts or diagrams, and sentiment analysis are out of scope for this beta.
Model Limitations: If you are currently using Gemini models, note that performance might degrade for very large files.
FAQs
How do I enable structured file analysis?
This should be turned on by default, unless you have requested to opt out of new features.
As a Glean admin, you can self-serve manage the list of users that have access by going to Admin Console > Platform > Assistant > Settings > Data analysis, and selecting the option that's right for you.
Refer to this how-to video on Data Analysis Admin Self-Serve.mov
Where do I follow up on the security whitepaper?
Here is the link to the security whitepaper.