You’ve built a model. It works well. Now you need it to actually do something.
Maybe predictions should feed into your CRM. Maybe they trigger alerts in your monitoring system. Maybe they power a customer-facing dashboard.
Kanva’s new REST API makes this integration straightforward.
From interface to integration
Kanva’s client app is great for building and understanding models. But production use often means connecting to other systems:
- Business applications that need predictions
- Scheduled jobs that run overnight
- Real-time systems that need instant responses
- Data pipelines that transform and route predictions
For these use cases, you need an API.
What’s available
The REST API provides programmatic access to your Kanva projects:
Predictions. Send input data, get predictions back. Works for all project types—forecasting, classification, regression.
Synchronous and asynchronous modes. Need instant responses? Use sync mode. Have large batches? Use async mode with webhooks.
Model checkpoints. Request predictions from specific model versions, not just the current one.
Project metadata. Query project configurations, feature schemas, and available checkpoints.
All authenticated with API keys you generate and manage.
Getting started quickly
We’ve added code snippet generation to make integration faster. In any project, click “Use API” to see:
Python examples. Copy-paste code for making predictions with the requests library.
cURL commands. Test from your terminal immediately.
Jupyter notebooks. Download a ready-to-run notebook with your project’s configuration.
The snippets use your actual project ID, feature names, and API endpoint. They’re not generic examples—they’re working code for your specific model.
Sync vs. async predictions
Two modes for different needs:
Synchronous. Send a request, wait for the response. Best for real-time applications where you need predictions immediately. Typical response times are under a second.
Asynchronous. Send a request, get a job ID, receive results via webhook. Best for large batches or when you don’t want to hold connections open.
Choose based on your use case. Many integrations use sync for real-time needs and async for batch processing.
Webhooks
For async predictions and job notifications, configure webhooks:
- Register your endpoint URL in project settings
- Kanva sends a POST when jobs complete
- Your system processes the results
Webhook payloads include the job ID, status, and prediction results. Secrets are encrypted for security.
API documentation
Full API documentation is available at api-ref.kanva.human-driven.ai. You’ll find:
- Complete endpoint reference
- Request and response schemas
- Authentication details
- Error codes and handling
- Interactive API explorer
The docs are generated from our OpenAPI specification, so they’re always current.
Example: CRM integration
Here’s a concrete example. Say you have a churn prediction model and want to flag high-risk customers in your CRM.
1. Generate an API key in Kanva project settings
2. Set up a scheduled job in your system (cron, Airflow, whatever you use)
3. The job queries your CRM for customers to score
4. Sends features to Kanva API:
import requests
response = requests.post(
"https://your-hub/api/v1/projects/{id}/predict",
headers={"Authorization": "Bearer YOUR_API_KEY"},
json={"inputs": customer_features}
)
predictions = response.json()
5. Updates CRM records with churn scores
6. Sales team sees flags in their normal workflow
The model you built in Kanva is now part of your business process.
What this enables
With API access, your models can:
- Power real-time applications
- Feed dashboards and reports
- Trigger automated workflows
- Integrate with any system that speaks HTTP
- Run in scheduled batch processes
The work you do in Kanva doesn’t stay in Kanva. It connects to your broader technology ecosystem.
The REST API is available now. Full documentation at api-ref.kanva.human-driven.ai. Questions? Reach out at hello@human-driven.ai