Integration Models Overview

6 ways to integrate PraisonAI recipes into your applications

Six integration models cover embedded SDK, CLI, sidecar HTTP, remote runner, events, and plugins.

python

from praisonaiagents import Agent

agent = Agent(
    name="Model Picker",
    instructions="Summarise PraisonAI recipe integration options.",
)
agent.start("Compare CLI versus embedded SDK for our CI pipeline.")

The user selects a tab or card below to compare latency, complexity, and setup time.

Diagram

PraisonAI recipes can be integrated into your applications using six distinct models. Each model has specific use cases, trade-offs, and implementation patterns.

Quick Comparison

By Latency

Model	Latency	Best For
Embedded SDK	Lowest	Python apps, notebooks
CLI Invocation	Low	Scripts, CI/CD
Plugin Mode	Low	IDE/CMS extensions
Local HTTP Sidecar	Medium	Microservices, polyglot
Remote Managed Runner	Medium	Multi-tenant, cloud
Event-Driven	Variable	Async workflows

By Complexity

Model	Complexity	Setup Time
Embedded SDK	Low	Minutes
CLI Invocation	Low	Minutes
Plugin Mode	Medium	Hours
Local HTTP Sidecar	Medium	Hours
Remote Managed Runner	High	Days
Event-Driven	High	Days

Local vs Remote

Model	Deployment	Network Required
Embedded SDK	In-process	No
CLI Invocation	Local subprocess	No
Plugin Mode	Local/Remote	Optional
Local HTTP Sidecar	Local container	Localhost only
Remote Managed Runner	Cloud/Self-hosted	Yes
Event-Driven	Distributed	Yes

Decision Guide

Diagram

Choose Your Integration Model

Model 1: Embedded SDK

Lowest latency – Direct Python integration with zero network overhead. Best for Python applications and notebooks.

Model 2: CLI Invocation

Language-agnostic – Invoke recipes from any language via subprocess. Perfect for scripts and CI/CD pipelines.

Model 3: Local HTTP Sidecar

Polyglot friendly – HTTP API running locally. Ideal for microservices and non-Python applications.

Model 4: Remote Managed Runner

Production-ready – Centralized, authenticated recipe execution. Built for multi-tenant cloud deployments.

Model 5: Event-Driven

Async at scale – Queue-based invocation for high-volume batch processing and decoupled architectures.

Model 6: Plugin Mode

Native UX – Embed recipes into IDEs, CMS platforms, and chat applications.

Feature Matrix

Feature	SDK	CLI	Sidecar	Remote	Event	Plugin
Streaming	✅	✅	✅	✅	❌	✅
Multi-tenant	❌	❌	❌	✅	✅	❌
Auth built-in	❌	❌	✅	✅	✅	❌
Async native	✅	❌	✅	✅	✅	✅
Process isolation	❌	✅	✅	✅	✅	✅
Hot reload	❌	✅	✅	✅	❌	❌

Quick Start

Simple Usage

Run a recipe from Python with the embedded SDK:

python

from praisonai import recipe

result = recipe.run("my-recipe", input={"query": "Hello"})
print(result.output)

With Configuration

Use CLI or a local HTTP sidecar when you need language-agnostic or service-style invocation:

CLI

praisonai recipe run my-recipe --input '{"query": "Hello"}' --json

HTTP Sidecar

# Start server
praisonai serve recipe --port 8765

# Invoke
curl -X POST http://localhost:8765/v1/recipes/run \
  -H "Content-Type: application/json" \
  -d '{"recipe": "my-recipe", "input": {"query": "Hello"}}'

Next Steps

Choose your integration model based on the decision guide above
Follow the dedicated page for step-by-step setup
Review Personas for role-specific guidance
Explore Use Cases for implementation patterns