Event-driven AI for backend frameworks

Receive AI-generated webhooks, execute custom business logic, and call back to ModelRiver: step-by-step guides for every major backend framework.

Overview

Every event-driven AI implementation follows the same pattern, regardless of framework:

Register a webhook endpoint in ModelRiver that points to your backend
Verify the signature on every incoming webhook
Process the AI response: run custom logic (database, APIs, validation)
Call back to the callback_url with your enriched data

The guides below show idiomatic implementations for each framework, including signature verification, async processing, error handling, and callback patterns.

Supported frameworks

Framework	Language	Highlights	Guide
Next.js	TypeScript	API routes, server actions, edge functions	View guide →
Nuxt.js	TypeScript	Server routes, Nitro engine, auto-imports	View guide →
Django	Python	Views, Celery tasks, Django REST Framework	View guide →
FastAPI	Python	Async handlers, BackgroundTasks, Pydantic	View guide →
Laravel	PHP	Queued jobs, middleware, event broadcasting	View guide →
Rails	Ruby	Active Job, Action Controller, credentials	View guide →
Phoenix	Elixir	GenServer, channels, Oban background jobs	View guide →
Spring Boot	Java	RestController, async processing, WebClient	View guide →
.NET	C#	Minimal APIs, hosted services, HttpClient	View guide →

Common pattern

Every framework implementation follows this structure:

POST /webhooks/modelriver  ←  ModelRiver delivers AI result
  │
  ├─ Verify mr-signature header
  ├─ Check type === "task.ai_generated"
  ├─ Return 200 immediately (acknowledge receipt)
  │
  └─ Background process:
       ├─ Extract ai_response, event, callback_url, customer_data
       ├─ Execute your custom business logic
       └─ POST enriched data to callback_url

Next steps

Serverless Databases: Supabase, PlanetScale, Neon, Convex
Webhooks reference: Signature verification, retries, and delivery
Event-driven AI overview: Architecture and flow