Introduction
Predict business outcomes and forecast time series with the REST API, Python, JavaScript or an AI agent.
Spavik turns prepared business data into predictions. Evaluate a table before training a model, or backtest a time series before forecasting its next steps. Evaluation and training are free; predictions and forecasts consume credits.
Make your first prediction
Set up an API key, evaluate your data and train a model.
Prepare and evaluate data
Choose the target, prevent leakage and select an honest evaluation protocol.
Python SDK
Work with CSV, JSON, pandas and synchronous or asynchronous clients.
TypeScript SDK
Use the typed client in Node.js and supported edge runtimes.
Connect an agent
Twelve MCP tools, with costs announced before each billable call.
REST API reference
Explore requests, responses and examples from the OpenAPI contract.
Two workflows
| Goal | Workflow | Billing |
|---|---|---|
| Predict a label or quantity from features | Evaluate → train → predict | Per scored row at the model engine's price |
| Forecast future series values | Backtest → forecast | Horizon per call, all series included |
For a reusable model, keep each input row and prediction, including its version and engine. Report the features and observed outcome later to update your model; include the saved prediction metadata to measure performance. See the feedback loop.
How is this guide?
Last updated on