Ch 9 — Framework Landscape
High-Level Overview
8 Steps
1
Click Next to explore the full landscape of agentic AI frameworks, protocols, and how to choose the right one for your use case.
Landscape
Why So Many Frameworks?
Different philosophies for the same problem
1
precision_manufacturing
Graph-First
LangGraph — explicit nodes, edges, state
vs
groups
Role-First
CrewAI — agents with roles, goals, backstories
vs
chat
Chat-First
AutoGen — agents as conversational participants
2
hub LangGraph & CrewAI & AutoGen — the big three in detail
Big Three
LangGraph vs CrewAI vs AutoGen
The most widely adopted frameworks
schema
LangGraph
Graph-based, max control, steep learning curve
vs
diversity_3
CrewAI
Role-based crews, fast setup, production Flows
vs
forum
AutoGen
Conversational agents, GroupChat, code exec
Vendor SDKs
OpenAI Agents SDK & Google ADK
First-party frameworks from model providers
3
smart_toy
OpenAI Agents SDK
Runner loop, handoffs, guardrails, low friction
vs
cloud
Google ADK
12 building blocks, Gemini-native, sessions
vs
window
Semantic Kernel
Microsoft, C#/Python, plugins, enterprise
4
bolt Lightweight & opinionated alternatives
Lightweight
Pydantic AI, smolagents, Agno, Mastra
Focused, opinionated, or minimal-abstraction tools
verified
Pydantic AI
Type-safe agents, DI, structured output, OTEL
compress
smolagents
~1K lines, CodeAgent, HuggingFace Hub
speed
Agno
High-perf runtime, AgentOS, ex-Phidata
javascript
Mastra
TypeScript-first, workflows, by Gatsby team
Protocols
MCP & A2A — Interoperability Standards
Framework-agnostic protocols for tools and agents
5
electrical_services
MCP
Model Context Protocol — universal tool access
+
swap_horiz
A2A
Agent-to-Agent — cross-system agent comms
=
public
Interop Layer
Any framework can talk to any tool or agent
6
chat_bubble Chat UIs — Chainlit, Streamlit, Gradio
Ecosystem
Chat UIs & Observability
The supporting tools around any framework
chat
Chainlit
Python chat UI, streaming, steps, auth
dashboard
Streamlit
Rapid prototyping, st.chat_message
widgets
Gradio
ML demos, ChatInterface, HF Spaces
Matrix
Side-by-Side Comparison
Key dimensions across all frameworks
7
compare
Control Level
Low-level graph vs high-level roles
+
lock_open
Vendor Lock-in
Model-agnostic vs single-provider
+
rocket_launch
Production Ready
Checkpoints, HITL, observability, scale
Guide
How to Choose
Decision framework for picking the right tool
8
help
Your Use Case
Simple chain? Multi-agent? Enterprise?
→
tune
Your Priorities
Control? Speed? Type safety? Ecosystem?
→
check_circle
Best Fit
Match framework to requirements