Skip to content

What is Myra AI Workspace

Myra AI Workspace (AIWS) is a secure AI work environment that combines the security and governance of an AI gateway with the productivity and adoption tools of an AI adoption platform — so every employee can work with AI safely, without compromising sensitive data. Under the hood it is a multi-tenant reverse proxy built upon the Global Myra Security CDN: it sits between your applications and upstream AI provider application programming interfaces (APIs), enforcing policy, managing credentials, and recording every request — all in-process, with no external sidecar required.

💡 Note: All policy enforcement — authentication, rate limiting, caching, and security checks — runs in-process with no additional network round trips, ensuring consistent low-latency enforcement at scale.

Key features

  • 21 provider integrations — OpenAI, Anthropic, Google Gemini, Vertex AI, AWS Bedrock, Azure OpenAI, Mistral, Groq, Together AI, Fireworks, Cerebras, DeepSeek, OpenRouter, Perplexity, SambaNova, xAI, NVIDIA NIM, Cloudflare Workers AI, Cohere, HuggingFace, Myra
  • Unified OpenAI-compatible endpoint — send any model name to /compat/chat/completions; the gateway resolves the provider automatically
  • Exact-match response caching — repeated identical prompts are served from cache, reducing cost and latency
  • Rate limiting — per-gateway or per-token request caps, enforced before any upstream call is made
  • Budget enforcement — hard spending caps at the gateway level and per auth token
  • Guardrail pipeline — a two-tier system: fast in-process regex/keyword checks (Tier 1) followed by optional HTTP sidecars — NLP PII Detector, Prompt Guard, and PII Protector (Tier 2)
  • BYOK key vault — provider API keys encrypted at rest with AES-256
  • Routing rules — an ordered rule engine that rewrites provider, model, and attaches fallback chains based on request metadata
  • Prometheus metrics — request counters and latency histograms with provider, tenant, status, and cache labels; exposed at /metrics
  • Admin UI and Playground — a web-based interface for managing tenants, gateways, users, and routing rules, and for running side-by-side multi-model comparisons

Use cases

Centralised API key management The gateway stores all provider API keys in one encrypted vault. Applications authenticate to the gateway with a scoped token and never access raw provider credentials directly.

Cost control and attribution Every request is logged with token counts, cost in USD, and a meta map of custom headers. Budget caps prevent runaway spend at both the gateway and per-user-token level.

Compliance and security policy enforcement The guardrail pipeline (NLP PII Detector, Prompt Guard, regex, keyword) applies PII scrubbing and content moderation centrally, before prompts reach a provider.

A/B routing and model experimentation Routing rules redirect a configurable percentage of traffic to a different provider or model, with automatic fallback if the primary fails. The Playground presents results side-by-side for comparison.

Self-hosted model routing Requests route to a self-hosted, OpenAI-compatible endpoint via provider_base_urls. Application code remains identical whether it targets a self-hosted model or any cloud provider.

Architecture overview

flowchart TD
    Client(["Consumer<br/>POST /v1/{tenant}/{gateway}/..."])
    Client --> A

    subgraph GW ["Myra AI Workspace"]
        A["Access phase<br/>auth · rate-limit · IP allowlist"]
        B["Content phase<br/>cache · guardrails · routing<br/>BYOK · cost · cache-store"]
        C["Log phase<br/>structured log · Prometheus"]
        A --> B --> C
    end

    B --> UP["Upstream Providers<br/>21 total"]
    C --> DB[("Config Store<br/>& Request Logs")]
Capability Description
Platform Built upon the Global Myra Security CDN
Providers 21 AI providers via a unified API
Security Authentication, rate limiting, budget enforcement, and guardrail pipeline (regex, NLP PII Detector, Prompt Guard, PII Protector) — all enforced in-process
Encryption Provider API keys encrypted at rest with AES-256
Observability Structured request logs, metrics, real-time admin dashboard
Admin UI Web-based admin interface with playground for model testing

See also