PageMantle Company Dataset
PageMantle is a fictional company that sells software for building and hosting websites. This synthetic dataset by Deepnote contains its conversations, documents, customer records, billing and analytics.
The dataset imitates the data found in services such as Slack, Notion and Stripe. You work with local files, without connecting to the real services.
Use it to study how AI agents find and connect information across company tools. For example, an agent could investigate a customer problem using Slack, Linear and Notion.
Explore a customer handoff across Slack, Linear and Notion in Deepnote, with the sample data, code and results already included.
What's included
Records are saved as of September 30, 2026, including September activity and older business records from eleven services:
- Work: Slack, Linear (issue tracking), Notion and Google Calendar.
- Customers and billing: HubSpot and Stripe.
- Analytics: PostHog, Google Analytics, Google Search Console, BigQuery and Snowflake.
Formats include SQLite, Parquet, Markdown, JSON and PDF. The data guide explains the files and how to open them.
Size: 5.97 GB compressed, 10.11 GB extracted, 101,337 files.
Start small
To run the example locally, use sample/ and examples/ from the release ZIP and open the quickstart notebook. These folders contain about 1.3 MB. See the setup instructions.
Or explore Linear directly with these four files from the release, arranged as shown:
download/
βββ data/
β βββ linear.tar.gz # Linear data, 344 KB
βββ manifest.json # Archive verification details
βββ file-index.jsonl.gz # Individual file checks
βββ tools/
βββ verify.py # Checks your download
With Python 3 installed, run these commands in a terminal from download/ to check and extract the archive:
python3 tools/verify.py --service linear
mkdir -p unpacked
tar -xzf data/linear.tar.gz -C unpacked
Show five issue identifiers and titles with Python 3. No extra packages are needed:
python3 - <<'PY'
import sqlite3
with sqlite3.connect("file:unpacked/linear/linear.sqlite?mode=ro", uri=True) as db:
for row in db.execute("SELECT identifier, title FROM issues ORDER BY identifier LIMIT 5"):
print(*row, sep=": ")
PY
For all services, use the eleven archives in data/, run python3 tools/verify.py, and extract each into unpacked/. See the verification guide for other check options.
License and citation
Use and adapt the data under CC BY 4.0, including commercially, with attribution and change notices.
Suggested citation: Deepnote. PageMantle Company Dataset. Version 1.0.0. DOI: 10.5281/zenodo.24005613. See CITATION.cff and release changes.
- Downloads last month
- 142