Data policy and privacy | ATX AnalyticsTable Docs

Data policy and privacyWhat may leave the machine: data levels and per-class ceilings, the payload preview, PII masking and its limits, saved connections, and what is and is not sent. Every payload bound for a model — from Ask Greg, from AI_PROMPT and from the AI analysis step — passes through one gate in the backend. That gate reduces the data to what the policy allows and then masks what is left. The policy is read from the installation's configuration, never from the conversation, so nothing typed into the chat or found in a cell can widen it.

Where to find it

Ask Greg (beta) → Governance → Data policy
Ask Greg (beta) → Grounding → What it knows
Ask Greg (beta) → Grounding → Index project

What this page covers

Data levels
The Data policy dialog
The payload preview
Masking personal data
Saved connections
What is sent, and what is not

Settings and controls documented

schema-only
schema+aggregates
schema+samples
full
Your level
Ceilings by endpoint class
Sample row limit
Mask personal data before sending
Masked columns
Blocked columns
Show the payload before the first question of a session
Let the agent read from saved connections
One AI answer per person
Extra rules for the agent
Show the agent's rules
Your question and the conversation so far in this session (masked)
The agent's scope rules and any administrator rules
A description of the open project: sheet names and column headers, the project database's tables and their columns, saved and pinned queries, parameters, and existing objects by name and kind
Passages retrieved from the manual and from this project's index
Results of the queries and range reads the agent runs, reduced to your level and masked

Full page

The full page — procedures and the explanation of every setting — is available to signed-in users.

Glossary terms on this page

PII masking: definition

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.