Scaling
Vantage is designed with a multi-tenant architecture that supports growing teams, increasing data volumes, and expanding usage patterns.
Multi-Tenant Isolation
Each company operates in an isolated data environment. This means:
- No data bleed — Company A's dashboards, workflows, and credentials are invisible to Company B
- Independent scaling — Each company's usage is tracked and billed separately
- Per-company AI context — AI responses are personalized per company without cross-contamination
Organizations add a layer above companies for centralized management without breaking isolation.
Token-Based Usage Metering
All operations in Vantage are metered using a token-based system:
| Operation Type | Token Cost |
|---|---|
| AI features (summaries, insights, queries) | Varies by provider and model |
| Workflow executions | Based on node count and data volume |
| Integration data syncs | Based on data transferred |
| Dashboard data operations | Based on query complexity |
Billing Modes
| Mode | How It Works |
|---|---|
| Prepaid | Purchase tokens upfront. Operations deduct from your balance. |
| Pay-As-You-Go | Usage is tracked and billed at the end of each billing cycle. |
Token usage is tracked per-category and per-operation in Settings → Account → Usage & Tokens. Organization-level usage logs aggregate consumption across all companies.
AI Provider Scaling
Vantage supports 7 AI providers, allowing you to choose based on performance, cost, and capability:
| Provider | Use Case |
|---|---|
| Intuidy AI | Default, no setup required |
| OpenAI | GPT-4o and variants for general-purpose AI |
| Claude | Anthropic models for nuanced analysis |
| Gemini | Google models for multimodal tasks |
| DeepSeek | Cost-effective for high-volume workloads |
| Grok | xAI models |
| Mistral | European AI models |
AI Query Settings
For large datasets, you can toggle Process Large Datasets in Settings:
- Enabled — AI processes all rows (slower but more accurate)
- Disabled — AI samples data (faster but may miss edge cases)
Dashboard Performance
Dashboard tiles are designed for efficient data rendering:
- Each tile independently manages its data source and refresh cycle
- Grid layout uses responsive design that adapts to screen size
- Drag-and-drop tile repositioning uses optimized re-rendering
- AI summaries are generated asynchronously and cached
Workflow Execution
Workflows scale through:
- Scheduled execution — Trigger workflows on cron-based schedules (every N minutes, daily, weekly)
- Batch processing — AI nodes support configurable batch sizes for processing large datasets
- Node-level execution — Each node runs independently, and you can monitor status per-node (pending, running, complete, error)
- Nested workflows — Use Execute Workflow and Run Workflow nodes to compose smaller workflows into larger pipelines
Integration Connections
Integration connections are managed per-company:
- Each company maintains its own set of credentials
- OAuth tokens are refreshed automatically when possible
- Database connections support standard connection parameters (host, port, database, credentials)
- Connection health is displayed on the Integrations page