AI runtime infrastructure
Frequently asked questions
Answers about Zyntry AI runtimes, model routing, private data, integrations, security, observability, and launch access.
Frequently asked questions
What is Zyntry?
Zyntry is AI runtime infrastructure for building and operating AI-powered applications. It provides one governed layer for models, routing, knowledge, integrations, security, and observability.
What is an AI runtime?
An AI runtime is the execution layer behind an AI feature. It receives a request, applies context and policy, selects an eligible model or provider, runs approved tools, and returns a response with safe evidence and telemetry.
Do I need to choose a model?
No. Automatic routing can select an eligible model based on request intent, capabilities, latency, cost, and provider health. You can also configure an explicit provider and model when your application requires it.
Can I connect private company data?
Yes. Runtimes can use private uploaded documents and approved connections. Project boundaries, source policies, retrieval settings, and credentials remain under your control.
Can I use my own integrations?
Yes. Custom integrations can describe an approved API contract, select read or write capabilities, configure authentication, test the connection, and attach only the capabilities a runtime needs.
How does Zyntry keep AI requests safe?
Runtimes are read-only by default, credentials stay vaulted, write actions can require confirmation, and requests can be protected with prompt-injection checks, PII redaction, rate limits, IP rules, and audit events.
Can Zyntry work with an existing application?
Yes. Keep your product code focused on the user experience and send server-side requests to the stable invoke endpoint with a scoped Bearer API key and runtime_id.
How does launch access work?
Eligible automatic-routing requests can use platform-managed Gemini for seven days from the first eligible request. Fair-use request and token limits apply, and usage is visible in Runtime Console.