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.
Step - / 8
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
1
Detail
close