Air-gapped, on-premise
COVE
Generative MC/DC unit tests for safety-critical flight code.
Air-gapped, on site Generative MC/DC unit tests Requirements conscious
Menu
Air-gapped
Nothing leaves your control
COVE and Compleye both run inside your facility, on hardware we supply and administer, used day to day by your own engineers.
There is no cloud tier, no telemetry and no phone-home. Your data stays under your control, inside your facility.
Your facility, our appliance - Our hardware appliance is installed in your facility. You control access.
No network egress - Neither product has a connection to the internet or to any external network interface.
Predictable operational costs - No AI cloud token fees. Processing runs on the appliance in your facility rather than in a metered service.
What it buys you
Software Certification Metrics Improvements
Accelerate schedule - Significant improvement in unit test development and execution compared to manual labor intensive methods. Requirements improvement and gap closure become an explicit part of the process rather than a discovery made near the end.
Preserve budget - Coverage closure is where safety-critical programs lose money late, when there is least room to absorb it.
Certification evidence - AI-generated unit tests, separately validated by Verity Gates. Automated reports provide coverage records and other software artifacts for your certification package.
The problem
The rule is not the risk. The closure loop is.
NASA NPR 7150.2D requires 100% modified condition/decision coverage for identified safety-critical components. Meeting it is where programs lose schedule.
78.5%
of experienced Level A respondents rated MC/DC moderately or extremely difficult
74.5%
rated its cost substantial or nearly prohibitive
Both figures are from the NASA and FAA Streamlining Software Aspects of Certification survey of experienced Level A practitioners.
Products
Two products, one discipline
Binary and object-code analysis
Compleye
No source code required.
Air-gapped, on site Certify a binary is not malicious Scans against cyber vulnerabilities such as CWE and CVE
How we make it trustworthy
Verity Gates
Verity Gates are used to satisfy deterministic authorized outputs. These guard rails ensure output is accurate and verified.
The Verity Gate runs outside of the LLMs, containing data and action integrity.
Truthfulness of the output is ensured by deterministic checks and guard rails that run outside the model.
The platform
What the tooling does
Staff, industry experience, and AI-enabled tools for safety-critical code AI-generated and separately validated unit tests with 100% code coverage Requirements development, improvement and gap closure as an explicit part of the process Complexity identification, with recommended code changes for reducing complexity to meet certification requirements Cyber vulnerabilities and identification of malicious algorithms, from source code or object code Liability reduction in critical civil and DoD applications Significant cost savings from high code coverage and an accelerated schedule
Where it applies
Built for human-rated software first
-
Human-rated spaceflight Human-rated software for space is the initial use case. -
DoD mission-critical Mission-critical software benefits from faster schedule and lower code complexity even where human-rated requirements are not explicitly required. -
Critical civil applications Liability reduction in all critical civil and DoD applications.