What is Vaikora?
Vaikora is an AI runtime control gateway that enforces deterministic policy on every AI agent action before it executes. Open-core: MIT-licensed gateway, commercial Control Plane with audit chain, compliance presets, and SLA.
How is Vaikora different from a guardrails library?
Guardrails libraries run callbacks around an LLM call. Vaikora enforces deterministic policy on the proposed action and produces a tamper-proof audit receipt. The decision is signed into a SHA-256 append-only chain, replayable by auditors without vendor cooperation.
Which LLM providers does Vaikora support?
First-class adapters for OpenAI, Anthropic, Google Gemini, and OpenRouter at the LLM level. Agent-to-agent (A2A) and Model Context Protocol (MCP) at the protocol level. Adding a new provider is a configuration change.
What compliance frameworks are covered?
SOC 2 Type II, HIPAA, GDPR, PCI DSS, and ISO 27001 ship as pre-built presets. The presets handle logging, evidence collection, and policy-mapping requirements out of the box.
How fast is the policy engine?
Decisions land in under 500ms at the p95 against the production Control Plane. Self-hosted gateway latency depends on your infrastructure; typical pilot deployments measure single-digit milliseconds for the engine itself.
Where can I buy Vaikora?
AWS Marketplace SaaS (three Vaikora connectors live as of 2026), Azure Sentinel (Vaikora-AzureSecurityCenter solution), the Anthropic MCP registry (vaikora-guard-mcp), and direct procurement.