| What it does | Security runtime for autonomous AI agents and multi-agent workflows
| Know exactly where every AI dollar goes
|
|---|
| Stage | MVP
| Not given
|
|---|
| Founded | Not given
| Not given
|
|---|
| Based in | Not given
| PY
|
|---|
| Pricing model | Free trial
| Freemium
|
|---|
| Pricing | Starter: $29/mo with 7-day free trial, 250,000 requests/mo, 1,500 mins sandbox compute, 5 custom Wasm rules. Pro: $79/mo, 1,000,000 requests/mo, 5,000 mins sandbox compute, unlimited custom Wasm policies. Enterprise: Custom pricing. Pay-as-you-go overages: $8 per 100k extra requests and $0.005 per m
| Free: $0/month (1 developer, 1 project, 7-day event history). Starter: $10/month (3 developers, 2 projects, 30-day event history). Team: $50/month (10 developers, 5 projects, 90-day event history). Business: $100/month (100 developers, 100 projects, 180-day event history).
|
|---|
| Key features | - L4/L7 Proxy Interceptor for inline A2A and MCP protocol interception
- Sub-millisecond Wasm guardrails and PII redaction under 1.2ms
- Financial circuit breakers and spend locks with multi-node Redlock consensus
- Firecracker MicroVM sandboxing for untrusted code execution
- Human-in-the-Loop asynchronous state hydration for approval workflows
- Slack integration for approval requests and notifications
- Zero agent refactoring with drop-in integration
- Support for LangGraph, CrewAI, AutoGen, and Claude frameworks
- 99.99% uptime SLA for data plane proxy on Pro and Enterprise tiers
| - Track spending by project, developer, model and provider
- Privacy-preserving SDK that never collects prompts or responses
- Fail-open design ensures AI calls work even if Tokmetry API is down
- Support for OpenAI, Anthropic and Google Gemini
- Cost calculator to estimate spending on different models
- No latency added to AI calls
- Historical cost tracking with versioned pricing
|
|---|
| What makes it different | - Infrastructure-level security enforcement outside agent context prevents prompt injection bypasses
- Sub-5ms p99 latency overhead with fail-closed guarantees
- Memory-safe Rust implementation with WebAssembly policy execution
- Zero payload retention and data sovereignty with SAML/OIDC support
| - Designed for privacy with only metadata collection, never storing prompts or API keys
- Fail-open architecture ensures your AI application keeps running even if the service goes down
- One-line integration with no proxy or gateway required
- Unified view across multiple AI providers with normalized costs
- Waste detection and cost comparison between models
|
|---|
| Community votes | 0 | 0 |
|---|
| DR (Domain Rating by Ahrefs) | Not checked yet ● no change since last check | Not checked yet ● no change since last check |
|---|
| Trust Flow (Majestic) | 0 | 0 |
|---|
| Citation Flow (Majestic) | 0 | 22 |
|---|
| Referring domains (Majestic) | 0 | 2 |
|---|
Tokmetry