Problem

Consus addresses the bottleneck that keeps US defense and aerospace contractors from using frontier AI on sensitive work. ITAR, CUI, FedRAMP, and related data-boundary requirements turn each model or provider choice into an integration and compliance project: teams must keep export-controlled and proprietary data in authorized US environments, produce assessor-ready evidence, and repeatedly review new models. For contractors with bespoke mission workflows and few spare software engineers, that paperwork and infrastructure burden slows AI adoption; a small team could otherwise ship a mission application in weeks.

The killer use case is an engineer moving among models while building something sensitive, such as Inversion Space’s autonomy, simulation, and software work on re-entry vehicles, without exposing proprietary work or starting a fresh security review each time. The economic pain is primarily the cost and delay of compliance engineering and model switching rather than raw token expense. Consus’s commercial model reflects this by charging a flat access fee plus model cost instead of seat licenses or token markup.

Product / Service

Consus provides compliant AI infrastructure through an API gateway. It accepts standard OpenAI-format requests, authenticates them, selects a backend based on the requested model and compliance level, and routes them through FedRAMP High, DoD IL2, or DoD IL5-authorized environments in AWS GovCloud, GCP Assured Workloads, and Azure Government. Customers can access Claude, Gemini, and GPT models through a common interface while prompts remain outside commercial cloud infrastructure and are not persisted. OpenAI SDK compatibility, audit-ready access logs, and a zero-persistence architecture reduce the need to rebuild applications as models change.

The delivery model is controlled B2B access for US-based defense contractors rather than self-serve software. Consus advertises a $25,000 access fee plus AI token cost for companies with fewer than 50 employees, enterprise pricing for larger organizations, no seats or minimums, and no token markup or per-developer licensing; access follows company review and a contract. The benefit is to make compliant frontier-model access a one-line integration while Consus carries the cloud, authorization, and evidence burden that would otherwise require months of engineering and compliance work.

Market

Consus operates in the secure and compliant AI infrastructure, AI gateway, and GovTech developer-tools market, focused narrowly on aerospace and defense contractors handling sensitive information. Its closest substitutes are government-cloud AI services and an internal build using AWS Bedrock in GovCloud, Azure Government, or Google Cloud Assured Workloads, as well as general-purpose AI gateway products such as Braintrust, Portkey, LiteLLM, Kong, SUSE, and Cloudflare. The evidence does not identify a single like-for-like direct competitor; Consus differentiates around the combination of model portability, US-only routing, ITAR/CUI/FedRAMP controls, and compliance evidence for contractors.

Public evidence indicates early commercial traction rather than demonstrated scale. Consus is listed by Y Combinator as an active Fall 2024 company founded in 2024, and its public case study identifies Inversion Space as a customer running GPT models in production through Consus’s compliant API; the case study also reports a production fix delivered in under a day. The available evidence does not establish revenue or justify labeling the company definitively pre-revenue, but it does show a live paid-access model, an active named customer deployment, and an early enterprise sales motion.