FAQ
Answers to common questions about Cortyxia.
Get clear answers about how Cortyxia handles AI memory, context retrieval, privacy, self-hosting, and cost reduction. Whether you are evaluating the platform for a team or deploying it across your organization, these questions cover the details that matter most.
Common Questions
Everything you need to know about the future of AI memory management.
Cortyxia operates as a high-performance proxy that works with any AI provider. By redirecting your application's base URL to your ISO endpoint, we intercept LLM calls in real-time. This allows us to perform semantic analysis, inject relevant context from memory, and optimize the request before routing it to your primary model provider.
Traditional AI interactions are isolated silos. Cortyxia consolidates every interaction across all your API keys and applications into a single, unified memory graph. This means knowledge gained by a developer in Cursor is immediately available to a support agent in Salesforce, creating a shared intelligence layer for the entire organization.
For organizations with strict security requirements, Cortyxia can be deployed entirely on-premise or within your VPC. The core proxy uses SQLite by default. PostgreSQL powers telemetry and analytics, and Redis is optional for caching. Your data never leaves your infrastructure.
Our retrieval engine is optimized for speed, typically adding less than 200ms to the total round-trip time. By using BM25 indexing with reranking, we ensure that the benefits of better context far outweigh the minimal latency overhead.
Yes. By pruning irrelevant context and optimizing prompts, we significantly reduce the total token count sent to the model. Most enterprise customers see a 40-60% reduction in token consumption while actually improving the quality and accuracy of model responses.