// Sheet 1 of 6
AI Chatbot & Agent Framework — Selection Guide
Covers: Copilot Studio • Dialogflow CX • Amazon Lex • LangChain/LangGraph • AutoGen • CrewAI
| Framework | Type | Open-Source? | Best Scenario(s) — Full Description | Primary Use Case | Multi-Agent? | LLM Flexibility | Memory / State | RAG Support | Voice / Telephony | Azure Native? | Deployment | Coding Required? | Learning Curve | Cost Model |
|---|
// Sheet 2 of 6
Decision Matrix
Score each criterion for your project. ✅ = Strong fit ⚠️ = Partial fit ❌ = Poor fit
| Decision Criterion | Copilot Studio |
Dialogflow CX |
Amazon Lex |
LangChain/ LangGraph |
AutoGen v0.4 |
CrewAI |
|---|
// Sheet 3 of 6
Real-World Use-Case Scenarios
Each scenario shows which framework to choose first (Primary), a fallback option (Secondary), and what to avoid.
| Scenario / Use Case | Industry | Primary Framework | Secondary Option | Avoid | Key Reason |
|---|
// Sheet 4 of 6
Quick Reference Card
IF your situation → THEN use this framework. Instant decision guide.
💡 Note: LangChain/LangGraph, AutoGen, and CrewAI are open-source and can run on ANY cloud or on-premise. Low-code platforms (Copilot Studio, Dialogflow CX, Lex) are vendor-managed SaaS with less flexibility but faster time-to-value.
// Sheet 5 of 6
Cross-Framework Cheat Sheet — Same Concept, Different Names
Every major AI chatbot / agent concept mapped across all 6 frameworks.
Amazon Lex
Copilot Studio
Dialogflow CX
LangChain / LangGraph
AutoGen v0.4
CrewAI
| Universal Concept | 🟡 Amazon Lex | 🔵 Copilot Studio | 🟢 Dialogflow CX | 🟣 LangChain / LangGraph | 🔴 AutoGen v0.4 | 🩵 CrewAI |
|---|
// Sheet 6 of 6
CORE RULE + Decision Tree (Fast Mental Model)
Cheat sheet for quick platform selection: deterministic vs reasoning vs multi-agent workflows.
CORE RULE (1-liner)
If the problem is deterministic → use Dialogflow / Lex
If the problem requires reasoning → use Autogen / CrewAI
⚡ DECISION TREE (Fast Mental Model)
1. Is it simple Q&A or intent matching?
→ YES → Dialogflow CX
2. Is it voice-first?
→ YES → Amazon Lex
3. Does it require multi-step reasoning / tools / workflows?
→ YES → Autogen
4. Does it involve multiple roles (planner, executor, reviewer)?
→ YES → CrewAI
🔍 RULES BY SYSTEM CHARACTERISTICS
1. Determinism vs Intelligence
System Type Use
Predictable, rule-based Dialogflow CX
Adaptive, reasoning-based Autogen / CrewAI
2. Interaction Modality
Modality Platform
Chat UI Dialogflow CX
Voice / call center Amazon Lex
Multi-channel + agents Hybrid
3. Complexity Threshold
Complexity Platform
Simple (FAQ, routing) Dialogflow CX
Medium (API calls, workflows) Dialogflow + backend
High (autonomous agents) Autogen / CrewAI
4. Orchestration Need
Need Platform
Single-step response Dialogflow
Multi-step pipeline Autogen
Multi-agent collaboration CrewAI
5. Data Dependency
Data Pattern Platform
Predefined answers Dialogflow CX
RAG / document search Autogen / CrewAI
Real-time DB queries Hybrid
🧠 INTUITION (Consultant-Level Insight)
Think in layers:
Dialogflow CX / Lex = Interface layer (NLU router)
Autogen / CrewAI = Reasoning & orchestration layer
👉 Most real systems = combination
🧪 REAL-WORLD MAPPING
Use Case Choice
FAQ chatbot Dialogflow CX
Call center voice bot Amazon Lex
AI research assistant Autogen
AI project manager agent CrewAI
Enterprise support system Dialogflow + Autogen
🚀 GOLDEN HEURISTIC (Use in Interviews)
“Start with the simplest deterministic system. Only introduce agents when you hit limits in reasoning, orchestration, or dynamic data handling.”
⚠️ COMMON MISTAKE
❌ Using Autogen/CrewAI for FAQ
→ leads to:
higher latency
higher cost
unpredictable answers
✅ FINAL CHEAT SHEET
FAQ / Intent → Dialogflow CX
Voice → Amazon Lex
Reasoning → Autogen
Multi-agent → CrewAI
If you want, I can next:
Build a real client scenario → platform selection breakdown
Or give you a system design template (interview-ready)