AI memory layer / personal data connector / knowledge graph API

Turn chats, documents and notes into user-portable, authoritatively callable AI memory layers

After users import personal data, a personal knowledge graph with source references is generated, which can be retrieved and called by different AI applications through APIs within the scope of user authorization.

01
Access personal dataUsers upload or paste chat records, documents, and notes, and the system creates data source records.
02
Build a memory graphThe system generates node, edge, and source references within the data scope confirmed by the user.
03
Configure API callsThe user configures the authorization scope for an application name and generates call instructions, sample responses and audit rules on this site.
04
Auditing and Portable ExportLog access events, revocation status, and generate machine-readable export package descriptions.

Data source ready0 data sourcesCount the number of memory_sources with import_status=ready

Number of graph nodes0 nodesCount the number of nodes in memory_graph.node_list under the current owner_user_id

Number of authorized applications0 appsCount the number of api_access_packages with access_status=active

Access audit eventsNo access record yetCount the number of events in audit_export_report.audit_events

TRACEABLE OUTPUT

personal memory space

Results, sources, failures, approvals, and rollbacks are saved by project.

0 live runs

∅

No business results yet

No chats, documents or notes have been imported yet.

PRIVACY

Data use boundary

The chat records, documents, and notes you upload or paste may contain personal identity, preferences, contacts, work content, and other sensitive information. The system is only used to generate data source records, knowledge graphs, authorization packages and audit reports within this site; unauthorized content will not be sent to external production systems. Current 10 Role Review Supplement: Default Data Minimization: Only the minimal fields required to complete imports, graphs, authorizations, audits, and exports are saved. ; Derived fields may also be sensitive; content_digest, source_ref, node_list, edge_list, source_refs, sample_response, audit_events, export_url, schema_summary all require access control and minimal display. ;The audit log does not record the complete raw content, query or response by default. ;Deletion should overwrite the available status of original files, pasted text, derived maps, sample responses, and export links; auditing retains only minimal security fields. ; The model must have user confirmation before processing personal data, minimize the sending field, and record prompt_version, model, and run_id; the output must not write the ready status if the schema verification fails.

RETENTION

Data retention

The source records, map versions, authorization packages and audit reports in the current memory space are retained by default until the user deletes or chooses to clean them after exporting; after deletion, the records in this site should enter an unavailable state, and the minimum audit records can retain the operation time, operator and deletion events according to security requirements.

SAFETY

Human responsibility and rollback

This product processes personal data actively provided by users and generates knowledge graphs, API authorization instructions and audit reports. The generated results may contain extraction errors, omissions, conflicts or over-generalization; when sensitive personal information, trade secrets or third-party data are involved, the user must confirm the scope of authorization before use.

Rolling back only means restoring the import records, graph version, API authorization package status or report status in this site to the previous version, or marking it as revoked/archived; it does not mean that any unconnected external AI applications, third-party data sources or production systems have been rolled back.