AI Decision Graph
Decision-first AI concepts

AI Decision Graph

Make better AI decisions. Understand when to use RAG, agents, fine-tuning, MCP, workflows, prompt engineering, and the concepts around them.

Start with AI Agents Choose RAG or Fine-tuning Browse Decision Paths Explore Graph
local graph preview RAG neighborhood
RAG Embedding Vector Database Chunking Reranking Hybrid Search Fine-tuning Long Context
Popular paths

Start from a product decision.

Agent products

Build an AI Agent

Understand when autonomy is useful, and when a deterministic workflow is safer.

 Integration

Understand MCP

Decide whether you need a protocol layer or simple tool/function calling.

 MVP baseline

Start with Prompt Engineering

Use prompts first when the model already knows enough and the task boundary is still changing.
Featured comparisons

High-intent AI choices.

Agent vs Automation

Agent vs Automation: Known Path or Uncertain Path?

Automation follows a known path. Agents choose a path under uncertainty. Start with automation unless uncertainty is the product value.

 Agent vs Workflow

Agent vs Workflow: Start Controlled, Add Autonomy Later

Most teams should start with workflows, not agents. Use agents only when the next step genuinely depends on uncertain observations.

 MCP vs Function Calling

MCP vs Function Calling: Integration Boundary or In-App Tool?

MCP is not better function calling. Function calling is an in-app model interface; MCP is useful when tool and context integration becomes a reusable product boundary.

 MCP vs Tool Calling

MCP vs Tool Calling: Protocol Boundary or Model Capability?

Tool calling is the model asking to use a capability. MCP is a protocol boundary for exposing tools and context across clients.

 Prompt Engineering vs Fine-tuning

Prompt Engineering vs Fine-tuning: Which Should You Try First?

Start with prompt engineering while the task is changing. Consider fine-tuning when behavior is stable, repeated, and backed by examples.

 RAG vs Fine-tuning

RAG vs Fine-tuning: Which Should You Choose?

Choose RAG for private or changing knowledge. Choose fine-tuning for repeated behavior, style, or task consistency.

 RAG vs Long Context

RAG vs Long Context: Does Long Context Replace Retrieval?

Long context can replace retrieval setup for bounded inputs. It does not replace retrieval strategy when the corpus grows, repeats, needs ranking, or needs source control.

 Vector Database vs Full-text Search

Vector Database vs Full-text Search: Which Retrieval Should You Use?

Use vector search for semantic similarity, full-text search for exact terms, and hybrid search when users need both.
Core concepts

The first decision nodes.

rag medium

RAG

Use RAG when the model needs private, updated, source-backed, or domain-specific knowledge.
agents high

Agent

Use an agent when the task needs adaptive planning across uncertain steps; use a workflow when the path is known.
engineering low

Prompt Engineering

Use prompt engineering first when the model already has enough knowledge and you need to clarify task boundaries, format, tone, or reasoning steps.
agents medium

MCP

Use MCP when reusable tool and context integrations matter across clients; use direct function calling for a narrow in-app MVP.
rag high

Fine-tuning

Use fine-tuning when you need consistent behavior across many similar tasks, not when you only need new facts.
rag medium

Vector Database

Use a vector database when semantic retrieval needs to scale beyond simple in-memory search.