We physically cannot see your data
When you run a query, it goes directly from your machine to your databases — it does not pass through our servers. This isn't a policy promise; it's a consequence of how the product is built. Queries, results, and credentials stay on your device by default.
Read-only by design: every query you run executes as SELECT only — the query engine rejects INSERT, UPDATE, DELETE, and any DDL (DROP, ALTER, CREATE, TRUNCATE, MERGE) outright, so a mistyped or overly broad query can never change or delete production data. Writing data anywhere is a separate, explicit action: you choose a destination and DataIncisive appends your query results there via INSERT — it never runs UPDATE or DELETE, and never edits or removes an existing row as a side effect of just running a query. For editing existing rows, keep using your usual SQL client.
We do operate servers for account login, licensing, and billing — for example, verifying your subscription when you upgrade. Those servers never see your data.
On Enterprise, if you run a self-hosted Control Plane (for SSO, row-level security, centralized credential management, or audit logging), it runs entirely on your own infrastructure, not ours. It sends us one lightweight, anonymous heartbeat a day — install ID, organization name, and active user count, nothing else — so we have basic adoption visibility, and it's non-blocking: login and queries keep working even if it fails. Separately, if you turn on audit logging, the full text of each query is sent to that same self-hosted Control Plane — never to us — so your own team has a queryable record of who ran what.
Nothing sensitive is ever stored in the clear
Database credentials
Encrypted at rest (AES-256-GCM) using a key generated and stored only on your device. Never uploaded anywhere. Exporting a configuration file to share with a colleague automatically strips credentials out first.
Query-result spill files
Large result sets that spill to a temporary file on disk are encrypted at rest too — not just held safely in memory and left exposed the moment they're written down.
Centrally-managed credentials Enterprise
Encrypted at rest on your Control Plane and issued to a connecting client just-in-time. A developer running a query never sees, types, or stores the actual password, key, or connection string.
Governance that follows your org, not a spreadsheet
Single Sign-On
SAML 2.0 and generic OIDC, so access follows your existing identity provider.
Role-based access control
Admin and Member roles, enforced at every point a query, source registration, or export happens.
Row-level security
A signed, per-user query predicate resolved from your Control Plane and enforced on every query — with genuine native-database enforcement as a second layer on Postgres, SQL Server, Oracle, and Elasticsearch.
Column-level security
Per-role column access control — a denied column's value never leaves the source database, on any connector.
SCIM provisioning
Automatic, instant deprovisioning the moment someone leaves your identity provider's roster.
A record of who ran what, and when
Every query, save, and export is logged with the full SQL text, the identity that ran it, and the result — matching how enterprise audit trails work elsewhere (Snowflake query history, pgAudit), not just anonymized metadata. Logs are written locally first and forwarded to your own Control Plane; a slow or unreachable Control Plane never blocks a query.
How we keep our own house in order
We run regular internal security reviews of our own code and dependencies, and act on what we find — our most recent review was completed in September 2026. Both of our codebases run automated dependency-vulnerability scanning on every change and on a weekly schedule, so a newly disclosed vulnerability in something we depend on gets caught automatically rather than discovered by accident.
Independent, third-party validation
SOC 2 and an independent penetration test are both on our roadmap as we scale. Neither is complete today — we'd rather tell you that plainly than overstate where we are. We'll update this page the moment either changes.
Found a security issue?
We take every report seriously and will work with you to understand and fix it.
Report a vulnerability