Infrastructure for model routing, traffic governance, and workflow delivery.
VectorDuct is an open-source middleware stack for engineering teams. Standardize foundation model schemas, enforce multi-tenant token quotas, and wire asynchronous generation pipelines into your existing tools—entirely within your own VPC.
{
"trace_id": "req_9f8a2c1e",
"tenant": "org_prod_01",
"auth": { "method": "bearer_rbac", "policy": "tier-2-prod", "verified": true },
"rate_limit": { "window": "1m", "remaining": 598, "limit": 600 },
"quota_budget_usd": { "used": 14.20, "ceiling": 250.00 }
}01 / STACK
Modular components with transparent upstream attribution.
vectorduct-gateway
Hardened API gateway distribution featuring multi-tenant RBAC, deterministic token budgets, automatic circuit breaking, and structured token telemetry sinks.
dsh-vertex-adapter
Zero-dependency protocol bridge translating OpenAI and Anthropic schemas into native Google Gemini API wire formats with strict BYOC isolation.
vectorduct-bot-bridge
Event-driven connectors bridging team messaging platforms and internal webhooks to asynchronous text and visual generation pipelines.
02 / RUNTIME
Zero-knowledge BYOC execution across cloud-native runtimes.
Zero-Knowledge BYOC Runtime
Credentials never leave your controlled execution boundary. Requests flow through vectorduct-gateway for RBAC enforcement and token telemetry before dsh-vertex-adapter normalizes upstream Google Gemini API payloads in memory.
Engineered for Cloud-Native Runtimes
Google Cloud (First-Class Support)
Native Google Gemini API protocol adaptation, stateless deployment on Cloud Run / GKE, and structured log exports to Cloud Logging & BigQuery.
Container & VPC Portable
Ships as lightweight Docker images ready for Kubernetes, AWS ECS, or bare-metal private VPCs.
curl -s https://localhost:3000/v1/chat/completions \
-H "Authorization: Bearer $VECTORDUCT_API_KEY" \
-H "X-BYOC-Provider: google-gemini" \
-d '{"model":"gemini-2.5-pro","messages":[{"role":"user","content":"ping"}]}'03 / DEPLOYMENT & COMMERCIAL EDITIONS
Open-source core. Enterprise-grade control plane and SLA.
Adopt our individual routing and gateway modules for free under permissive licenses, or partner with VectorDuct for managed cloud control planes, enterprise SSO/SAML, and dedicated VPC engineering.
For individual developers and small teams running single-node gateways and protocol adapters.
- Full source access to dsh-vertex-adapter, vectorduct-gateway, and vectorduct-bot-bridge
- Standard Bring-Your-Own-Credentials (BYOC) isolation
- Basic token quota management and local Docker Compose deployment
- Community GitHub issue support
For growing AI product teams requiring centralized telemetry, multi-region routing, and zero maintenance overhead.
- Everything in Community, plus Managed Control Plane with hybrid data-plane nodes
- Multi-region active-active failover across Google Gemini API, Google Cloud channels, and fallback providers
- Long-term token cost attribution, department chargeback reports, and BigQuery / Cloud Logging native exporters
- Automated upstream schema hot-patching & priority email/Slack engineering support
For organizations with strict data residency, custom multimodal pipelines, and high-concurrency SLA requirements.
- Private VPC Deployment: Terraform / GKE / Cloud Run fleet provisioning inside your own cloud account
- Enterprise Identity & Compliance: Okta / Azure AD SAML SSO, granular RBAC audit trails, and zero-retention compliance guarantees
- Custom Adapter & Workflow Engineering: Bespoke protocol bridges, internal IM bot integrations, and dedicated multimodal queues
- 99.95% Uptime SLA with dedicated systems architect support and 2-hour incident response