Eric, founder of Consus, has launched an LLM gateway built for companies handling sensitive data. Consus routes AI requests to ITAR-restricted and controlled unclassified information (CUI) compliant infrastructure. The company says it satisfies ITAR restrictions and recently completed a FedRAMP equivalency audit against the 410 FedRAMP High controls, the U.S. government's strictest security baseline for cloud services handling sensitive unclassified data.
Eric spent much of his career in aerospace and defense, launching national security payloads at SpaceX and working as an early employee at Castelion. He says compliance requirements often stop companies from adopting AI, even when they see its value. Some firms fear model providers will train on their data, violate ITAR rules, or lose business after a failed cyber assessment.
Consus works as a compliant inference layer that connects to tools such as ChatGPT, Claude, Open WebUI and the Pi coding agent, as well as custom applications. Every request passes through a single OpenAI-compatible endpoint, where keys are verified and policy is enforced. Model names must carry a compliance suffix, such as gpt-5.6-sol:fedramp-high, and bare model names are rejected. Customers can build model allowlists and set spend and model restrictions per user through the Consus Portal.
The company argues that few aerospace and defense firms use all three major GovClouds (AWS, Azure and Google), and that models are split across providers. Consus runs these environments behind one endpoint and handles the compliance work for each. It also states that it does not store customer data at rest, keeping prompts and completions in memory only for the length of each request.
Source: Hacker News · Summarized by HeadlinesBriefing