Thinking in LangGraph ↗
yesSummary: Learn how to think about building agents with LangGraph
Editorial Notes
LangGraph models agents as state machines with explicit nodes and edges, and this conceptual doc is the key to understanding why. Before writing any LangGraph code, internalize the idea that every agent decision point is a node, every transition is an edge, and all context lives in a typed state object that flows through the graph. This is a fundamentally different paradigm from CrewAI’s role-based orchestration or OpenAI’s handoff-based delegation — it gives you maximum control over execution flow but requires more upfront design. A common mistake is trying to build everything as a single monolithic graph; instead, compose smaller subgraphs for each capability.
Original Documentation
Documentation Index#
Fetch the complete documentation index at: https://docs.langchain.com/llms.txt Use this file to discover all available pages before exploring further.
Learn how to think about building agents with LangGraph
When you build an agent with LangGraph, you will first break it apart into discrete steps called nodes. Then, you will describe the different decisions and transitions from each of your nodes. Finally, you connect nodes together through a shared state that each node can read from and write to.
In this walkthrough, we’ll guide you through the thought process of building a customer support email agent with LangGraph.
Start with the process you want to automate#
Imagine that you need to build an AI agent that handles customer support emails. Your product team has given you these requirements:
The agent should:
- Read incoming customer emails
- Classify them by urgency and topic
- Search relevant documentation to answer questions
- Draft appropriate responses
- Escalate complex issues to human agents
- Schedule follow-ups when needed
Example scenarios to handle:
1. Simple product question: "How do I reset my password?"
2. Bug report: "The export feature crashes when I select PDF format"
3. Urgent billing issue: "I was charged twice for my subscription!"
4. Feature request: "Can you add dark mode to the mobile app?"
5. Complex technical issue: "Our API integration fails intermittently with 504 errors"To implement an agent in LangGraph, you will usually follow the same five steps.
Step 1: Map out your workflow as discrete steps#
Start by identifying the distinct steps in your process. Each step will become a node (a function that does one specific thing). Then, sketch how these steps connect to each other.
flowchart TD
A[START] --> B[Read Email]
B --> C[Classify Intent]
C -.-> D[Doc Search]
C -.-> E[Bug Track]
C -.-> F[Human Review]
D --> G[Draft Reply]
E --> G
F --> G
G -.-> H[Human Review]
G -.-> I[Send Reply]
H --> J[END]
I --> J[END]The arrows in this diagram show possible paths, but the actual decision of which path to take happens inside each node.
Now that we’ve identified the components in our workflow, let’s understand what each node needs to do:
Read Email: Extract and parse the email contentClassify Intent: Use an LLM to categorize urgency and topic, then route to appropriate actionDoc Search: Query your knowledge base for relevant informationBug Track: Create or update issue in tracking systemDraft Reply: Generate an appropriate responseHuman Review: Escalate to human agent for approval or handlingSend Reply: Dispatch the email response
Notice that some nodes make decisions about where to go next (Classify Intent, Draft Reply, Human Review), while others always proceed to the same next step (Read Email always goes to Classify Intent, Doc Search always goes to Draft Reply).
Step 2: Identify what each step needs to do#
For each node in your graph, determine what type of operation it represents and what context it needs to work properly.
Use when you need to understand, analyze, generate text, or make reasoning decisions
Use when you need to retrieve information from external sources
Use when you need to perform external actions
Use when you need human intervention
LLM steps#
When a step needs to understand, analyze, generate text, or make reasoning decisions:
Data steps#
When a step needs to retrieve information from external sources:
Action steps#
When a step needs to perform an external action:
User input steps#
When a step needs human intervention:
Step 3: Design your state#
State is the shared memory accessible to all nodes in your agent. Think of it as the notebook your agent uses to keep track of everything it learns and decides as it works through the process.
What belongs in state?#
Ask yourself these questions about each piece of data:
<span class=“card-start” data-card-raw=“title=“Include in state” icon=“check”"> Does it need to persist across steps? If yes, it goes in state.
<span class=“card-start” data-card-raw=“title=“Don’t store” icon=“code”"> Can you derive it from other data? If yes, compute it when needed instead of storing it in state.
For our email agent, we need to track:
- The original email and sender info (can’t reconstruct these later)
- Classification results (needed by multiple later/downstream nodes)
- Search results and customer data (expensive to re-fetch)
- The draft response (needs to persist through review)
- Execution metadata (for debugging and recovery)
Keep state raw, format prompts on-demand#
A key principle: your state should store raw data, not formatted text. Format prompts inside nodes when you need them.
This separation means:
- Different nodes can format the same data differently for their needs
- You can change prompt templates without modifying your state schema
- Debugging is clearer – you see exactly what data each node received
- Your agent can evolve without breaking existing state
Let’s define our state:
from typing import TypedDict, Literal
# Define the structure for email classification
class EmailClassification(TypedDict):
intent: Literal["question", "bug", "billing", "feature", "complex"]
urgency: Literal["low", "medium", "high", "critical"]
topic: str
summary: str
class EmailAgentState(TypedDict):
# Raw email data
email_content: str
sender_email: str
email_id: str
# Classification result
classification: EmailClassification | None
# Raw search/API results
search_results: list[str] | None # List of raw document chunks
customer_history: dict | None # Raw customer data from CRM
# Generated content
draft_response: str | None
messages: list[str] | NoneNotice that the state contains only raw data – no prompt templates, no formatted strings, no instructions. The classification output is stored as a single dictionary, straight from the LLM.
Step 4: Build your nodes#
Now we implement each step as a function. A node in LangGraph is just a Python function that takes the current state and returns updates to it.
Handle errors appropriately#
Different errors need different handling strategies:
| Error Type | Who Fixes It | Strategy | When to Use |
|---|---|---|---|
| Transient errors (network issues, rate limits) | System (automatic) | Retry policy | Temporary failures that usually resolve on retry |
| LLM-recoverable errors (tool failures, parsing issues) | LLM | Store error in state and loop back | LLM can see the error and adjust its approach |
| User-fixable errors (missing information, unclear instructions) | Human | Pause with interrupt() | Need user input to proceed |
| Unexpected errors | Developer | Let them bubble up | Unknown issues that need debugging |
Add a retry policy to automatically retry network issues and rate limits:
from langgraph.types import RetryPolicy
workflow.add_node(
"search_documentation",
search_documentation,
retry_policy=RetryPolicy(max_attempts=3, initial_interval=1.0)
)
```
<span class="tab-end"></span>
<span class="tab-start" data-tab-title="LLM-recoverable"></span>
Store the error in state and loop back so the LLM can see what went wrong and try again:
```python
from langgraph.types import Command
def execute_tool(state: State) -> Command[Literal["agent", "execute_tool"]]:
try:
result = run_tool(state['tool_call'])
return Command(update={"tool_result": result}, goto="agent")
except ToolError as e:
# Let the LLM see what went wrong and try again
return Command(
update={"tool_result": f"Tool error: {str(e)}"},
goto="agent"
)
```
<span class="tab-end"></span>
<span class="tab-start" data-tab-title="User-fixable"></span>
Pause and collect information from the user when needed (like account IDs, order numbers, or clarifications):
```python
from langgraph.types import Command
def lookup_customer_history(state: State) -> Command[Literal["draft_response"]]:
if not state.get('customer_id'):
user_input = interrupt({
"message": "Customer ID needed",
"request": "Please provide the customer's account ID to look up their subscription history"
})
return Command(
update={"customer_id": user_input['customer_id']},
goto="lookup_customer_history"
)
# Now proceed with the lookup
customer_data = fetch_customer_history(state['customer_id'])
return Command(update={"customer_history": customer_data}, goto="draft_response")
```
<span class="tab-end"></span>
<span class="tab-start" data-tab-title="Unexpected"></span>
Let them bubble up for debugging. Don't catch what you can't handle:
```python
def send_reply(state: EmailAgentState):
try:
email_service.send(state["draft_response"])
except Exception:
raise # Surface unexpected errors
```
<span class="tab-end"></span>
<span class="tab-group-end"></span>
### Implementing our email agent nodes
We'll implement each node as a simple function. Remember: nodes take state, do work, and return updates.
<AccordionGroup>
<Accordion title="Read and classify nodes" icon="brain">
```python
from typing import Literal
from langgraph.graph import StateGraph, START, END
from langgraph.types import interrupt, Command, RetryPolicy
from langchain_openai import ChatOpenAI
from langchain.messages import HumanMessage
llm = ChatOpenAI(model="gpt-5-nano")
def read_email(state: EmailAgentState) -> dict:
"""Extract and parse email content"""
# In production, this would connect to your email service
return {
"messages": [HumanMessage(content=f"Processing email: {state['email_content']}")]
}
def classify_intent(state: EmailAgentState) -> Command[Literal["search_documentation", "human_review", "draft_response", "bug_tracking"]]:
"""Use LLM to classify email intent and urgency, then route accordingly"""
# Create structured LLM that returns EmailClassification dict
structured_llm = llm.with_structured_output(EmailClassification)
# Format the prompt on-demand, not stored in state
classification_prompt = f"""
Analyze this customer email and classify it:
Email: {state['email_content']}
From: {state['sender_email']}
Provide classification including intent, urgency, topic, and summary.
"""
# Get structured response directly as dict
classification = structured_llm.invoke(classification_prompt)
# Determine next node based on classification
if classification['intent'] == 'billing' or classification['urgency'] == 'critical':
goto = "human_review"
elif classification['intent'] in ['question', 'feature']:
goto = "search_documentation"
elif classification['intent'] == 'bug':
goto = "bug_tracking"
else:
goto = "draft_response"
# Store classification as a single dict in state
return Command(
update={"classification": classification},
goto=goto
)
```
</Accordion>
<Accordion title="Search and tracking nodes" icon="database">
```python
def search_documentation(state: EmailAgentState) -> Command[Literal["draft_response"]]:
"""Search knowledge base for relevant information"""
# Build search query from classification
classification = state.get('classification', {})
query = f"{classification.get('intent', '')} {classification.get('topic', '')}"
try:
# Implement your search logic here
# Store raw search results, not formatted text
search_results = [
"Reset password via Settings > Security > Change Password",
"Password must be at least 12 characters",
"Include uppercase, lowercase, numbers, and symbols"
]
except SearchAPIError as e:
# For recoverable search errors, store error and continue
search_results = [f"Search temporarily unavailable: {str(e)}"]
return Command(
update={"search_results": search_results}, # Store raw results or error
goto="draft_response"
)
def bug_tracking(state: EmailAgentState) -> Command[Literal["draft_response"]]:
"""Create or update bug tracking ticket"""
# Create ticket in your bug tracking system
ticket_id = "BUG-12345" # Would be created via API
return Command(
update={
"search_results": [f"Bug ticket {ticket_id} created"],
"current_step": "bug_tracked"
},
goto="draft_response"
)
```
</Accordion>
<Accordion title="Response nodes" icon="edit">
```python
def draft_response(state: EmailAgentState) -> Command[Literal["human_review", "send_reply"]]:
"""Generate response using context and route based on quality"""
classification = state.get('classification', {})
# Format context from raw state data on-demand
context_sections = []
if state.get('search_results'):
# Format search results for the prompt
formatted_docs = "\n".join([f"- {doc}" for doc in state['search_results']])
context_sections.append(f"Relevant documentation:\n{formatted_docs}")
if state.get('customer_history'):
# Format customer data for the prompt
context_sections.append(f"Customer tier: {state['customer_history'].get('tier', 'standard')}")
# Build the prompt with formatted context
draft_prompt = f"""
Draft a response to this customer email:
{state['email_content']}
Email intent: {classification.get('intent', 'unknown')}
Urgency level: {classification.get('urgency', 'medium')}
{chr(10).join(context_sections)}
Guidelines:
- Be professional and helpful
- Address their specific concern
- Use the provided documentation when relevant
"""
response = llm.invoke(draft_prompt)
# Determine if human review needed based on urgency and intent
needs_review = (
classification.get('urgency') in ['high', 'critical'] or
classification.get('intent') == 'complex'
)
# Route to appropriate next node
goto = "human_review" if needs_review else "send_reply"
return Command(
update={"draft_response": response.content}, # Store only the raw response
goto=goto
)
def human_review(state: EmailAgentState) -> Command[Literal["send_reply", END]]:
"""Pause for human review using interrupt and route based on decision"""
classification = state.get('classification', {})
# interrupt() must come first - any code before it will re-run on resume
human_decision = interrupt({
"email_id": state.get('email_id',''),
"original_email": state.get('email_content',''),
"draft_response": state.get('draft_response',''),
"urgency": classification.get('urgency'),
"intent": classification.get('intent'),
"action": "Please review and approve/edit this response"
})
# Now process the human's decision
if human_decision.get("approved"):
return Command(
update={"draft_response": human_decision.get("edited_response", state.get('draft_response',''))},
goto="send_reply"
)
else:
# Rejection means human will handle directly
return Command(update={}, goto=END)
def send_reply(state: EmailAgentState) -> dict:
"""Send the email response"""
# Integrate with email service
print(f"Sending reply: {state['draft_response'][:100]}...")
return {}
```
</Accordion>
</AccordionGroup>
## Step 5: Wire it together
Now we connect our nodes into a working graph. Since our nodes handle their own routing decisions, we only need a few essential edges.
To enable [human-in-the-loop](/oss/python/langgraph/interrupts) with `interrupt()`, we need to compile with a [checkpointer](/oss/python/langgraph/persistence) to save state between runs:
<Accordion title="Graph compilation code" icon="sitemap" defaultOpen={true}>
```python
from langgraph.checkpoint.memory import MemorySaver
from langgraph.types import RetryPolicy
# Create the graph
workflow = StateGraph(EmailAgentState)
# Add nodes with appropriate error handling
workflow.add_node("read_email", read_email)
workflow.add_node("classify_intent", classify_intent)
# Add retry policy for nodes that might have transient failures
workflow.add_node(
"search_documentation",
search_documentation,
retry_policy=RetryPolicy(max_attempts=3)
)
workflow.add_node("bug_tracking", bug_tracking)
workflow.add_node("draft_response", draft_response)
workflow.add_node("human_review", human_review)
workflow.add_node("send_reply", send_reply)
# Add only the essential edges
workflow.add_edge(START, "read_email")
workflow.add_edge("read_email", "classify_intent")
workflow.add_edge("send_reply", END)
# Compile with checkpointer for persistence, in case run graph with Local_Server --> Please compile without checkpointer
memory = MemorySaver()
app = workflow.compile(checkpointer=memory)The graph structure is minimal because routing happens inside nodes through Command objects. Each node declares where it can go using type hints like Command[Literal["node1", "node2"]], making the flow explicit and traceable.
Try out your agent#
Let’s run our agent with an urgent billing issue that needs human review:
Run with a thread_id for persistence#
config = {“configurable”: {“thread_id”: “customer_123”}} result = app.invoke(initial_state, config)
The graph will pause at human_review#
print(f"human review interrupt:{result[’interrupt’]}”)
When ready, provide human input to resume#
from langgraph.types import Command
human_response = Command( resume={ “approved”: True, “edited_response”: “We sincerely apologize for the double charge. I’ve initiated an immediate refund…” } )
Resume execution#
final_result = app.invoke(human_response, config) print(f"Email sent successfully!”)
</Accordion>
The graph pauses when it hits `interrupt()`, saves everything to the checkpointer, and waits. It can resume days later, picking up exactly where it left off. The `thread_id` ensures all state for this conversation is preserved together.
## Summary and next steps
### Key Insights
Building this email agent has shown us the LangGraph way of thinking:
<span class="card-group-start" data-cols="2"></span>
<span class="card-start" data-card-title="Break into discrete steps" data-card-icon="sitemap" data-card-href="#step-1-map-out-your-workflow-as-discrete-steps"></span>
Each node does one thing well. This decomposition enables streaming progress updates, durable execution that can pause and resume, and clear debugging since you can inspect state between steps.
<span class="card-end"></span>
<span class="card-start" data-card-title="State is shared memory" data-card-icon="database" data-card-href="#step-3-design-your-state"></span>
Store raw data, not formatted text. This lets different nodes use the same information in different ways.
<span class="card-end"></span>
<span class="card-start" data-card-title="Nodes are functions" data-card-icon="code" data-card-href="#step-4-build-your-nodes"></span>
They take state, do work, and return updates. When they need to make routing decisions, they specify both the state updates and the next destination.
<span class="card-end"></span>
<span class="card-start" data-card-title="Errors are part of the flow" data-card-icon="alert-triangle" data-card-href="#handle-errors-appropriately"></span>
Transient failures get retries, LLM-recoverable errors loop back with context, user-fixable problems pause for input, and unexpected errors bubble up for debugging.
<span class="card-end"></span>
<span class="card-start" data-card-title="Human input is first-class" data-card-icon="user" data-card-href="/oss/python/langgraph/interrupts"></span>
The `interrupt()` function pauses execution indefinitely, saves all state, and resumes exactly where it left off when you provide input. When combined with other operations in a node, it must come first.
<span class="card-end"></span>
<span class="card-start" data-card-title="Graph structure emerges naturally" data-card-icon="sitemap" data-card-href="#step-5-wire-it-together"></span>
You define the essential connections, and your nodes handle their own routing logic. This keeps control flow explicit and traceable - you can always understand what your agent will do next by looking at the current node.
<span class="card-end"></span>
<span class="card-group-end"></span>
### Advanced considerations
<Accordion title="Node granularity trade-offs" icon="adjustments">
<span class="callout-start" data-callout-type="info"></span>
This section explores the trade-offs in node granularity design. Most applications can skip this and use the patterns shown above.
<span class="callout-end"></span>
You might wonder: why not combine `Read Email` and `Classify Intent` into one node?
Or why separate Doc Search from Draft Reply?
The answer involves trade-offs between resilience and observability.
**The resilience consideration:** LangGraph's [durable execution](/oss/python/langgraph/durable-execution) creates checkpoints at node boundaries. When a workflow resumes after an interruption or failure, it starts from the beginning of the node where execution stopped. Smaller nodes mean more frequent checkpoints, which means less work to repeat if something goes wrong. If you combine multiple operations into one large node, a failure near the end means re-executing everything from the start of that node.
Why we chose this breakdown for the email agent:
* **Isolation of external services:** Doc Search and Bug Track are separate nodes because they call external APIs. If the search service is slow or fails, we want to isolate that from the LLM calls. We can add retry policies to these specific nodes without affecting others.
* **Intermediate visibility:** Having `Classify Intent` as its own node lets us inspect what the LLM decided before taking action. This is valuable for debugging and monitoring—you can see exactly when and why the agent routes to human review.
* **Different failure modes:** LLM calls, database lookups, and email sending have different retry strategies. Separate nodes let you configure these independently.
* **Reusability and testing:** Smaller nodes are easier to test in isolation and reuse in other workflows.
A different valid approach: You could combine `Read Email` and `Classify Intent` into a single node. You'd lose the ability to inspect the raw email before classification and would repeat both operations on any failure in that node. For most applications, the observability and debugging benefits of separate nodes are worth the trade-off.
Application-level concerns: The caching discussion in Step 2 (whether to cache search results) is an application-level decision, not a LangGraph framework feature. You implement caching within your node functions based on your specific requirements—LangGraph doesn't prescribe this.
Performance considerations: More nodes doesn't mean slower execution. LangGraph writes checkpoints in the background by default ([async durability mode](/oss/python/langgraph/durable-execution#durability-modes)), so your graph continues running without waiting for checkpoints to complete. This means you get frequent checkpoints with minimal performance impact. You can adjust this behavior if needed—use `"exit"` mode to checkpoint only at completion, or `"sync"` mode to block execution until each checkpoint is written.
</Accordion>
### Where to go from here
This was an introduction to thinking about building agents with LangGraph. You can extend this foundation with:
<span class="card-group-start" data-cols="2"></span>
<span class="card-start" data-card-title="Human-in-the-loop patterns" data-card-icon="user-check" data-card-href="/oss/python/langgraph/interrupts"></span>
Learn how to add tool approval before execution, batch approval, and other patterns
<span class="card-end"></span>
<span class="card-start" data-card-title="Subgraphs" data-card-icon="hierarchy" data-card-href="/oss/python/langgraph/use-subgraphs"></span>
Create subgraphs for complex multi-step operations
<span class="card-end"></span>
<span class="card-start" data-card-title="Streaming" data-card-icon="broadcast" data-card-href="/oss/python/langgraph/streaming"></span>
Add streaming to show real-time progress to users
<span class="card-end"></span>
<span class="card-start" data-card-title="Observability" data-card-icon="chart-line" data-card-href="/oss/python/langgraph/observability"></span>
Add observability with LangSmith for debugging and monitoring
<span class="card-end"></span>
<span class="card-start" data-card-title="Tool Integration" data-card-icon="tool" data-card-href="/oss/python/langchain/tools"></span>
Integrate more tools for web search, database queries, and API calls
<span class="card-end"></span>
<span class="card-start" data-card-title="Retry Logic" data-card-icon="rotate" data-card-href="/oss/python/langgraph/use-graph-api#add-retry-policies"></span>
Implement retry logic with exponential backoff for failed operations
<span class="card-end"></span>
<span class="card-group-end"></span>
***
<span class="callout-start" data-callout-type="note"></span>
[Edit this page on GitHub](https://github.com/langchain-ai/docs/edit/main/src/oss/langgraph/thinking-in-langgraph.mdx) or [file an issue](https://github.com/langchain-ai/docs/issues/new/choose).
<span class="callout-end"></span>
<span class="callout-start" data-callout-type="note"></span>
[Connect these docs](/use-these-docs) to Claude, VSCode, and more via MCP for real-time answers.
<span class="callout-end"></span>Link last verified
June 7, 2026.
View original ↗
Source: LangChain Docs
Appears in Learning Paths
Link last verified: 2026-02-26