.// MODELS
Three models. One intelligent routing layer.
Vera routes inference requests to the optimal model based on task complexity, workspace policy, and data sensitivity requirements. Most requests never leave your infrastructure.
| Model | Role | Parameters | Hosting | Data Retention | Status |
|---|---|---|---|---|---|
| Qwen 3.5-9B | Primary inference | 9B | Self-hosted / Vera Cloud | Zero | Live |
| Claude API | Fallback (complex reasoning) | N/A (API) | Anthropic (zero-retention) | Zero | Live |
| Vera-Engine-9B | Enterprise operations (fine-tuned) | 9B | Self-hosted / Vera Cloud | Zero | Roadmap |
.// INFERENCE ARCHITECTURE
Intelligent routing. Zero wasted tokens.
Vera's model router evaluates each request and selects the optimal model based on task complexity, required capabilities, and your workspace's model policy. Simple queries stay on self-hosted infrastructure. Complex reasoning escalates when needed.
Input
Agent Request
Context + tools + constraints
Model Router
Complexity Assessment
Task analysis + policy check
Output
Validated Response
Safety filtered + audit logged
Standard Path (90%+ of requests)
Data queries, report generation, ticket triage, invoice processing, and standard workflow execution all route to Qwen 3.5-9B on self-hosted infrastructure. Zero external API calls. Full data sovereignty.
Complex Path (opt-in fallback)
Multi-step reasoning chains, complex code generation, nuanced natural language analysis, and creative tasks route to Claude API when enabled. Zero-retention guarantees. Admin-controlled per workspace.
.// DATA SOVEREIGNTY
Your data never trains. Your data never persists.
Zero data retention is not a policy toggle — it is an architectural guarantee. Inference data is processed and immediately discarded from model infrastructure. No conversation logs. No training data. No exceptions.
Self-Hosted = Zero Exposure
When running Qwen 3.5-9B on your infrastructure or Vera Cloud, data never leaves the processing environment. No external API calls. No network exposure.
API Fallback = Zero Retention
When Claude API is used for complex tasks, Anthropic's zero-retention API guarantees no data storage and no model training. Contractually enforced.
Audit Logs Stay In-Platform
Conversation history and audit trails are stored in your Vera OS database — governed by the same 1,607 RLS policies that protect all enterprise data.
Compliance-Ready
Architecture designed for SOC 2, GDPR, HIPAA, and ISO 27001 compliance. Data processing boundaries documented and auditable.
.// BRING YOUR OWN KEY
Your keys. Your contracts. Your control.
Bring Your Own Key (BYOK) support lets you use your own API keys and enterprise agreements with model providers. Route inference through your existing contracts for consolidated billing and compliance.
API Key Management
Store and manage API keys per workspace. Rotate keys without downtime. All keys encrypted at rest with tenant-specific encryption keys.
Contract Consolidation
Use your existing enterprise agreements with Anthropic, OpenAI, or other providers. API calls route through your accounts for consolidated billing.
Model Flexibility
Switch between model providers without changing agent configurations. Vera abstracts the model layer — your agents work with any compatible provider.
Models managed. Data sovereign.
Self-hosted inference with intelligent fallback routing and zero data retention.
.// READY TO DEPLOY?
Your competitors deployed AI agents last quarter. What's your timeline?
See how Vera puts AI agents into production across Finance, Sales, Support, HR, and Compliance — with governance your enterprise requires. Start with a 30-minute discovery call.
See how it works
Context Engine, Semantic Layer, and Action Engine — see the three-layer architecture that powers governed agent execution.
Explore the platform →From pilot to production in 4 weeks
In 30 minutes, describe your most painful workflow. Within 48 hours, receive a custom POC plan with ROI projections, integration requirements, and a deployment roadmap.
Book a discovery call →