Skip to content
Database integrations

Query operational evidence without losing control.

Connect approved views through narrowly scoped, read-oriented paths that preserve schema, query, and lineage context.

Evidence-grade intelligence · Human validation · Traceable decisions
Databases / evidence viewIn development
reproducible query path
Approved view · read-oriented
Fields constrainedRecords filteredTime range fixed
Snapshot retained for review
Illustrative view
  • ScopeApproved views only
  • QueriesRecorded and bounded
  • SchemaDrift made visible
  • SnapshotsReproducible evidence
In development

Database connectors are in development and cannot be enabled today.

This page describes the intake model we are building, not a connector you can switch on. No database provider is supported yet. If a specific system matters to your investigations, tell us — that input shapes what gets built first.

The ReturnValue approach

Controlled access is more useful than an unrestricted connection.

ReturnValue should know which records were requested, under what authority, from which schema version, and at what point in time before database evidence supports a financial finding.

01

Define the boundary

Use approved datasets, views, fields, filters, and time ranges aligned to a specific investigation purpose.

02

Validate the extract

Check types, keys, coverage, duplication, freshness, and unexpected schema changes before analysis.

03

Retain reproducibility

Record the source, query or extraction definition, snapshot time, and transformation path behind every accepted result.

What is being built

Real workflows decide what ships first.

Database connectors are in active development with design partners, and this page describes what is being built and how to influence the order. None are available yet, and none can be switched on today; this reflects what is in development, not what is shipped. If your team would benefit from one of these sources, we want to know, because demand from real teams sets the order.

Relational sources are first, and the connector design will extend to similarly shaped sources afterwards. Connectors will be read-oriented at the schema level to help prevent accidental source mutation, and connection credentials will be scoped to the smallest privilege the workflow needs. Tables and columns will land as first-class sources, with types preserved and lineage tracked.

A connector that returns rows is easy. One that holds up against schema drift, partial data, mixed encodings, and the authentication quirks of a real production warehouse is the actual work. We are doing that work with a small group of design partners before opening connectors to everyone. This page will say available when the connector clears that bar, not before.

Read-oriented by default
Connectors will be read-oriented at the schema level to help prevent accidental source mutation.
Scoped credentials
Connection credentials will be scoped to the smallest privilege the workflow needs.
Tenant isolation
Workspace isolation is designed to extend through the connector layer, so one workspace cannot reach the schema or data of another.
Audited access
Connector reads are planned to be recordable in the workspace audit trail alongside the rest of the work.
Connected investigation

Database evidence that can be reproduced and reviewed.

Reviewers can understand what was queried and why without receiving unrestricted access to the operational system.

  1. 01AuthorizeView approved
  2. 02QueryRequest bounded
  3. 03SnapshotEvidence fixed
  4. 04ValidateQuality tested
  5. 05TraceFinding linked
Evidence in context

Clarity at every decision point.

The safest useful connection is the narrowest one that can answer the financial question.

Illustrative view. Values shown are examples, not customer data.

Query to finding

Reproducible
  1. Approved viewread-oriented
  2. Bounded queryrecorded
  3. Findingsnapshot cited
See ReturnValue in your environment

Follow your first financial signal.

Databases — Integrations | ReturnValue.ai