LangChain integration

Secure your LangChain agents with AIM in 2 lines of code. Get complete audit trails, trust scoring, and security alerts for all tool invocations - without refactoring.

What you'll get

  • Secure existing LangChain agents (zero refactoring)
  • Complete audit trail of all tool uses
  • Dynamic trust scoring
  • Security alerts for anomalous behavior
  • Automatic action verification before tool execution

Integration time: 5 minutes

Code changes: 2-3 lines

Difficulty: Beginner

Quick start (5 minutes)

Step 1: Download SDK from AIM dashboard

Important: There is NO pip package for the AIM SDK. You must download it from your AIM dashboard with pre-configured credentials.

  1. 1. Login to AIM Dashboard → Settings → SDK Download
  2. 2. Click "Download Python SDK" (contains your credentials)
  3. 3. Extract the downloaded ZIP file
  4. 4. Install dependencies: pip install langchain langchain-openai keyring PyNaCl

Step 2: Register agent

In AIM dashboard (running on http://localhost:8080):

  1. Navigate to Agents → Register new agent
  2. Name: langchain-assistant
  3. Type: AI agent
  4. Copy the generated credentials
export AIM_URL="http://localhost:8080"
export AIM_AGENT_ID="your-agent-id"
export AIM_PRIVATE_KEY="your-private-key"
export OPENAI_API_KEY="your-openai-key"

Step 3: Add AIM to your LangChain agent

Before (unsecured)

from langchain.agents import AgentExecutor, create_openai_functions_agent
from langchain_openai import ChatOpenAI
from langchain.tools import Tool

# Your existing LangChain agent
llm = ChatOpenAI(model="gpt-4")
tools = [search_tool, calculator_tool]
agent = create_openai_functions_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools)

# Run agent (no security, no audit trail)
result = agent_executor.run("What's the weather in SF?")

After (secured with AIM) - just add 3 lines

from aim_sdk import secure  # ← Line 1: Import AIM
from aim_sdk.integrations.langchain import AIMCallbackHandler
from langchain.agents import AgentExecutor, create_openai_functions_agent
from langchain_openai import ChatOpenAI

# Register with AIM
aim_agent = secure("langchain-assistant")  # ← Line 2: Secure your agent

# Your existing LangChain agent (unchanged)
llm = ChatOpenAI(model="gpt-4")
agent = create_openai_functions_agent(llm, tools, prompt)

# Add AIM callback
agent_executor = AgentExecutor(
    agent=agent,
    tools=tools,
    callbacks=[AIMCallbackHandler(aim_agent=aim_agent)]  # ← Line 3: Add callback
)

# Run agent - now secured with full audit trail!
result = agent_executor.run("What's the weather in SF?")

That's it. Your LangChain agent is now secured with:

  • Complete audit trail of every tool invocation
  • Dynamic trust scoring (0-100%)
  • Security alerts for anomalous behavior
  • Automatic compliance reporting (SOC 2, HIPAA, GDPR)

What gets logged

Every tool invocation is logged to AIM with:

  • Tool name: Which tool was called (e.g., "search_database")
  • Arguments: What parameters were passed
  • Result: What the tool returned
  • Timestamp: When the action occurred
  • Agent ID: Which agent performed the action
  • Trust score: Agent's trust score at execution time
  • Verification status: Whether action was approved/denied

View in AIM dashboard

After running your LangChain agent, visit the AIM dashboard to see:

  • Activity feed: Stream of all tool invocations
  • Trust score timeline: How your agent's trust evolves over time
  • Security alerts: Any anomalous behavior detected
  • Audit logs: Complete compliance-ready audit trail
  • Performance metrics: Tool success rates, latency, errors

Advanced: explicit verification

For sensitive operations, you can require explicit verification before tool execution:

from aim_sdk.integrations.langchain import AIMVerifiedTool

# Wrap sensitive tools with verification
verified_db_tool = AIMVerifiedTool(
    aim_agent=aim_agent,
    tool=database_tool,
    require_verification=True  # Blocks until verified in AIM dashboard
)

# Add to your agent
tools = [search_tool, calculator_tool, verified_db_tool]

Next steps

Trust scoring

Learn how AIM calculates trust scores for your agents

Security alerts

Configure alerts for suspicious agent behavior

Compliance reporting

Generate SOC 2, HIPAA, GDPR compliance reports

API reference

Explore all 136 AIM API endpoints

Need help?

Questions about integrating AIM with LangChain? Support is available.