Safe by Design
Kyntic sits between your AI and your machines and is built to fail toward safety. If the device loses power or faults, a hardware relay brings the robot to a controlled stop rather than letting unchecked commands through. The safe default is always that nothing moves.
Two Isolated Checks
Every command passes two checks that run on separate hardware. An AI judge reviews intent against your rules. A deterministic safety core then verifies the actual motion against fixed physical limits. The deterministic core does not trust the AI judge or the upstream model, so a single compromised layer cannot let a dangerous command through.
The Deterministic Core Can't Be Talked Out Of It
The safety core never reads the AI's words or reasoning. It evaluates the physical command against hard mathematical bounds. Even a successful prompt injection or a fully compromised upstream model cannot move a motor past those bounds, because they are checked on outputs, not inputs.
Predictable, Auditable Decisions
Every block Kyntic makes is explainable and repeatable. The deterministic core gives the same answer for the same command every time. If we stop a command, we can show you exactly which limit it would have crossed.
Tamper-Evident Audit Logs
Kyntic keeps an append-only, cryptographically signed record of every command it sees and every one it blocks. These logs are designed to serve as audit-ready evidence for your safety and insurance reviews.
Hardened Appliance
Each box ships locked down: minimal attack surface, no inbound services exposed, tightly scoped communication, and isolation between the AI processing zone and the trusted safety core.
Strict Access Controls
Rule and limit changes are controlled and logged. Every appliance authenticates with unique credentials that can be revoked instantly, and sensitive secrets are held in dedicated encrypted storage.
Third-Party Validation
Kyntic is committed to independent review. Functional-safety assessment and penetration testing by accredited firms are planned, and results will be published on this page when available.
Standards We Build Around
Kyntic's checks are designed to map to the requirements of the following frameworks:
- ISO 10218 / ISO/TS 15066 -- Industrial and collaborative robot safety
- ISO 13849 / IEC 61508 -- Functional safety of control systems
- IEC 62443 -- Industrial automation and control systems security
Supply Chain Integrity
The software running on the appliance is built from a tightly controlled, minimal set of trusted dependencies, scanned for known vulnerabilities before every release, and produced through reproducible build pipelines.
Responsible Disclosure
If you discover a potential vulnerability, please report it to security@thekyntic.com or through our contact form. We commit to acknowledging all reports within 48 hours, providing status updates within 7 days, and crediting researchers (with their permission) after remediation. Kyntic will not pursue legal action against researchers who disclose vulnerabilities in good faith.