Constellation Gate AI

A drop-in proxy gateway that sits between AI agents and language models, screening every request and response for prompt injection attacks, secret leaks, and PII exposure while automatically reducing token spend through compression and caching. It sells to developers and operations teams building AI agents, from startups to enterprises. The product is positioned as an inline defense that enforces custom policies on every call, blocks attacks before they reach the model, and provides a tamper-evident audit log, outperforming the leading enterprise vendor on prompt-injection benchmarks.

Key features

• Blocks prompt injection attacks
• Redacts secrets and PII
• Stops indirect injection via tool calls
• Compresses outbound prompts to save tokens
• Prompt-cache injection for discounted rates
• Provides tamper-evident audit log
• Configurable policy for enforcement