# Make your app predict what happens next

> Turn historical data into predictions your app can use, with one API.

Source: https://spavik.co

## Your data already knows what happens next

If your app records an outcome, Spavik can predict it.

1. **Your app already records it.** Your database knows which customers cancelled, and what they looked like before.
2. **Spavik learns from it.** Point Spavik to that history and the column you want to predict. It first checks that your data is enough.
3. **Your app shows the answer.** Ask about any customer, current or new. The answer comes back to your app as a value you can display.

## SDKs, API and MCP

- **Python SDK** (`pip install spavik`): Train a model and get predictions from your Python scripts.
- **JavaScript SDK** (`npm install spavik`): Add predictions to your JavaScript or TypeScript backend.
- **REST API** (`https://docs.spavik.co/api-reference/openapi.json`): Call Spavik from any language. Import the OpenAPI contract into your HTTP client.
- **MCP** (`mcp.spavik.co`): Let your AI agents make predictions from structured data.

## Or let AI do the integration

Your coding assistant writes the calls. Your AI agent uses Spavik as a tool.

### With your coding assistant

Your assistant reads the Spavik documentation written for it, then adds the calls to your app.

For Claude or ChatGPT:

```text
I'm discovering Spavik. Read https://spavik.co/llms.txt and https://docs.spavik.co/llms.txt, and the pages they link to. Then tell me what Spavik is, its possible use cases, how I could integrate it into my existing app or an idea of an app I could build with it, which plans are available and how much it would cost me.
```

For Claude Code, Codex or Cursor:

```text
I'm discovering Spavik. Read https://spavik.co/llms.txt and https://docs.spavik.co/llms.txt, and the pages they link to. Then tell me what Spavik is, its possible use cases, how I could integrate it into this codebase based on the data it already records, or an idea of an app I could build with it, which plans are available and how much it would cost me.
```

### In your AI agent

Connect the hosted MCP server. Your agent calls Spavik on its own, on a schedule or when something happens, and acts on the predictions. Server: https://mcp.spavik.co. See https://spavik.co/mcp-integration.

## From data to predictions

Send Spavik data from your databases, business tools or files.

Predictions:

- Churn probability: 78%
- Conversion probability: 34%
- Estimated price: $1,285,000
- Delivery time: 31 hours
- Peak demand tomorrow: 1,040 MW
- Orders tomorrow: 1,640

## What happens to your data

- **One key, one workspace**: Each API key opens a single workspace. A key is shown once, stored as a hash and can be revoked at any time.
- **Encrypted archives**: Archived inputs, outputs and submitted outcomes are encrypted with AES-256-GCM, with a key for each model.
- **Deletion is final**: Archiving keeps a model and frees a slot. Deleting it purges it from the engine, with its pending outcomes.
- **No third-party AI**: Predictions are computed by Spavik’s own engine. Your data is not sent to other AI providers.

## Pricing

Credits pay for predictions. Your plan determines the available features and usage limits. See https://spavik.co/pricing.md for current plans.