Privacy policy
What this deployment receives, where data can go, how long known data classes remain, and how to request a change.
What enters Management Layer
These are categories, not a claim that every record exists for every user.
Management Layer processes data needed to answer requests, carry out enabled workflows, preserve business history, and diagnose the service. It should not receive unrelated personal data.
- Conversation and action content
- Prompts, form inputs, attachments, requested actions, and the answers or drafts produced from them.Provided by a user or created while carrying out that user's request.
- Identity and access facts
- Account identifiers, providers, routes, timestamps, and authentication or delivery outcomes used to protect and trace a session.Observed by the dashboard, API, ChatGPT surface, or connected transport.
- Business records
- Projects, people, mail, reports, risks, training, decisions, and evidence needed for management work and history.Entered in Management Layer or synchronized from an enabled business system.
- Connector payloads and receipts
- The request and response data needed to read from, write to, or confirm work in an enabled connector.Exchanged only when a workflow uses that connector.
Why it is used
The use must serve the management workflow or operation of this deployment.
- Answer management questions and produce briefings, summaries, and risk views.
- Execute requested workflows such as command intake, report ingestion, message drafting, and approved message sending.
- Maintain auditability, diagnose issues, and monitor service health and abuse.
Management Layer does not use operational data for unrelated consumer profiling. A new purpose should trigger a new policy review before data is reused.
Where data can go
Possible recipients depend on the workflow and which connectors are enabled.
Data is exchanged with a connected system only when an enabled workflow needs it. This public notice lists recipient categories; it does not claim that every connector is currently enabled or healthy.
- Business connectors
- Enabled Zoho services, Google Ads or Analytics, and configured internal business systems may receive the minimum data needed for a requested read, write, or synchronization.
- AI surfaces and providers
- ChatGPT and configured AI providers may process prompts, context, and outputs when the selected workflow invokes them.
- Service infrastructure
- Hosting, storage, logging, and reverse-proxy services process limited operational data to run, secure, and diagnose this deployment.
How long data stays
Unknown or external lifecycles are shown as such instead of being presented as deletion.
- General operational recordsNo fixed deadline
- This deployment does not currently configure one automatic deletion deadline for general management records. Treat them as retained for business continuity and review until an operator confirms removal.
- Inactive mail threadsArchive after 90 days
- A scheduled task archives inactive mail threads after 90 days. Archiving removes them from active views; it does not erase the underlying records.
- Google Ads marketing identifiersRedact after 30 days
- Raw click identifiers and hashed email or phone matching data used for revenue attribution are redacted at an absolute 30-day deadline. Non-identifying business facts may remain.
- Copies in connected systemsExternal lifecycle
- Disconnecting a connector stops future exchange but does not prove that the source system deleted its copy. The operator must confirm that separately with the connected provider.
Backups, logs, and provider-held copies do not have a single published duration in this notice. Treat them as retained until the operator confirms the applicable schedule and outcome.
Security and control
Access controls reduce exposure; they do not replace careful data choices.
Do not submit credentials or authentication codes in prompts, forms, attachments, or privacy-request email.
Request access, correction, or deletion
The deployment operator handles requests; this page does not create a ticket.
- Identify the account, workspace, or email address the request concerns.
- Name the data or connected system, the relevant time range, and whether you want access, correction, deletion, or disconnection.
- Do not include a password, authenticator code, API token, or other secret in the email.