Consus
An LLM gateway for enterprises handling sensitive data. Through a policy engine it provides data masking, multi-model routing and audit logs, helping regulated industries such as finance, healthcare and law use large language models while remaining compliant.
Timeline (last 90 days)
Consus is an LLM gateway incubated by YC F24 and designed specifically for enterprises handling sensitive data.
Consus scans the prompt before the request is sent, identifying sensitive content such as PII, keys, and internal codenames within it.
Consus allows enterprises to configure rules, choosing to block sensitive content, replace it with placeholders, or allow it to be sent while logging an audit trail.
Consus provides a unified API endpoint, so enterprises do not need to maintain different SDKs and authentication logic for each model vendor.
Consus supports routing by data sensitivity level: low-sensitivity requests go to public cloud models, while high-sensitivity requests are automatically routed to privately deployed models.
Consus records the initiator, data involved, routing path, and token consumption of every LLM call, forming an audit trail.
Consus's differentiation lies in treating "sensitive data handling" as a first-principles design goal rather than an after-the-fact add-on feature.
All Facts (7)
Consus is an LLM gateway incubated by YC F24 and designed specifically for enterprises handling sensitive data.
50%UnverifiedConsus scans the prompt before the request is sent, identifying sensitive content such as PII, keys, and internal codenames within it.
50%UnverifiedConsus allows enterprises to configure rules, choosing to block sensitive content, replace it with placeholders, or allow it to be sent while logging an audit trail.
50%UnverifiedConsus provides a unified API endpoint, so enterprises do not need to maintain different SDKs and authentication logic for each model vendor.
50%UnverifiedConsus supports routing by data sensitivity level: low-sensitivity requests go to public cloud models, while high-sensitivity requests are automatically routed to privately deployed models.
50%UnverifiedConsus records the initiator, data involved, routing path, and token consumption of every LLM call, forming an audit trail.
50%UnverifiedConsus's differentiation lies in treating "sensitive data handling" as a first-principles design goal rather than an after-the-fact add-on feature.
50%