Last reviewed: June 12, 2026
- PILLAR ONE
Security
How your data is protected. Authentication, encryption in transit and at rest, row-level isolation in the database, AI provider key isolation, and a security architecture built into the product from the first line of code. Honest about what is live today and what is still on our roadmap.
See the security posture → - PILLAR TWO
Privacy
How your data is handled. Your questions and analyses are encrypted at rest, isolated so only you can reach them, and never used to train AI. Operator access is redacted by default and every reveal is logged, the founder included. You can export or delete everything. The honest limits are stated plainly, not buried.
See what we protect → - PILLAR THREE
Honesty
How the analysis itself stays unbiased. The principles that govern every analysis Kindred produces: steelman every position, show the work, source everything, no political alignment, no nudging. Sourcing is held to enforced standards: reference lists built only from sources actually retrieved, citations that resolve to the real source, and a plainly stated limit on what is not yet verified. The pillar most products do not have, and the heart of what Kindred means by trust.
Read the principles →
Three pillars, one promise.
Most products treat security and privacy as a compliance checkbox, and trust as a tagline. Kindred treats trust as the whole product. A platform whose job is to help you reason clearly cannot ask you to take its reasoning on faith.
So we publish the security architecture in concrete technical detail, even where that means publishing what is not yet built. We publish the principles that govern every analysis, with the disciplined commitment to steelman every position rather than nudge you toward one. We publish a privacy policy designed for GDPR and CCPA and a true hard delete, because your data is yours.
If any of the three pillars fail, the others do not redeem them. A product that protects your data while subtly steering your conclusions is not trustworthy. A product that gives you fair analysis while leaking your data is not trustworthy. The commitment has to be all three, or it is none.