eqsiq.com

What It Is

Codename Ultraviolet（CN-UV） positions itself as “Governed AI with runtime proof” — a runtime governance and evidence system for enterprise generative AI applications. It does not provide foundation models itself; instead, it wraps a governance layer around the model inference boundary: without a compliant Governed Inference Envelope, calls to external models are not allowed. This approach suits enterprises that want to turn AI compliance from a documentation process into real-time control.

Core Capabilities

Its core mechanisms include the GIE header, which contains information such as Policy Pack, Identity Posture, Data tags, Lane ID, and Stop Profile, and is used to enforce policies when calling external models. The ΔV Lockstream Evidence Manifest hashes prompts and outputs and writes them into an immutable manifest, generating replayable evidence packages for auditing and root-cause analysis. The product also emphasizes Lane Portability, meaning the same policy pack can run across different model providers, reducing fragmentation of governance rules in multi-cloud environments.

API, Integrations, and Data Privacy

Public materials indicate that Ultraviolet will initially provide lightweight SDKs supporting governed model calls via Azure OpenAI, Amazon Bedrock, and Google Vertex. It leans toward a Microsoft-native architecture, built around Entra ID, Purview, and Teams workflows, and supports native exception handling and approvals in Teams. On data privacy, the product claims it can be deployed within the customer tenant, preserving data residency and identity systems; however, it has not disclosed encryption methods, data retention periods, compliance certifications, or details on third-party processing.

Pricing and Trial

The current page does not provide pricing, plans, free trial, or commercial contract information, offering only a “Become a design partner” option and a contact email address. This suggests it may still be in a design partner or early enterprise validation stage. Before procurement, buyers should focus on confirming available product versions, SLA, deployment timeline, support scope, and billing model.

Pros, Cons, and Best-Fit Users

Its strengths are a clear positioning and a focus on policy enforcement, audit evidence, cross-cloud portability, and Microsoft ecosystem integration for enterprise AI calls. It may appeal to financial services, healthcare, government-related organizations, and large enterprise IT governance teams. Limitations include sparse public information and a lack of customer cases, certifications, performance overhead data, policy configuration details, and Chinese-language support information. It is better suited to organizations that already use multi-cloud model calls, have strict compliance requirements, and maintain enterprise identity and approval workflows.

Access from China and Alternatives

Website accessibility from mainland China is unknown, and payment methods have not been disclosed. For deployment in China, enterprises should also verify network connectivity, cross-border data transfer requirements, availability of related Azure/AWS/Google services, and the contracting entity. Comparable options include Microsoft Purview, Azure AI governance capabilities, AWS Bedrock Guardrails, Google Vertex AI governance tools, as well as Credo AI, Fiddler AI, Lakera, and Protect AI.