CrewAI Embeddings

Embeddings represent text as vectors in a high dimensional space, placing similar meanings closer together. In CrewAI, they enable memory, helping agents retain context and recall past interactions.

• Convert text into high dimensional vector representations.
• Capture semantic similarity (e.g., “cat” and “kitten” are close).
• Enable agents to store and retrieve context through memory.
• Help connect past interactions with current tasks for better responses.

Role of Embeddings in CrewAI

Embeddings play a key role in enabling intelligent memory and information handling in CrewAI systems.

• Support contextual memory by helping agents recall relevant past information.
• Enable information retrieval through similarity based search.
• Allow cross session recall by storing embeddings in databases like ChromaDB.
• Ensure continuity, preventing agents from treating each interaction independently.

Implementation

This example shows how embeddings in CrewAI enable agents to store, retrieve and share context, with memory=True ensuring continuity across tasks.

• Agent: Defines the role, goal and backstory of an AI agent.
• Task: Specifies the task description, assigned agent and expected output.
• Context: Connects tasks so outputs from one task can be used in another.

import os
os.environ["OPENAI_API_KEY"] = "Your_API_Key"

from crewai import Agent, Task, Crew, Process

agent1 = Agent(role="Researcher", 
               goal="Research AI history", 
               backstory="An academic researcher.", 
               verbose=True)
               
agent2 = Agent(role="Writer", 
               goal="Write a summary", 
               backstory="A science writer.", 
               verbose=True)

task1 = Task(description="Collect facts about AI history.", 
             agent=agent1, 
             expected_output="A list of historical facts about AI.")
             
task2 = Task(description="Summarize the research into a short article.", 
             agent=agent2, 
             expected_output="A short article summarizing AI history.", 
             context=[task1])

1. Using OpenAI Embeddings

OpenAI embeddings can be used for memory, offering strong semantic similarity performance. Models like text-embedding-3-small are efficient, while larger variants provide higher accuracy.

• memory=True: Enables memory for the crew.
• embedder: Specifies the embedding provider and model.

crew = Crew(
    agents=[agent1, agent2],
    tasks=[task1, task2],
    process=Process.sequential,
    memory=True,
    embedder={
        "provider": "openai",
        "config": {"model": "text-embedding-3-small"}
    }
)

crew.kickoff()

Output:

2. Using Google Embeddings

Google embeddings are suitable for multilingual tasks and large-scale data. Models such as models/embedding-001 provide strong cross-lingual support.

• provider="google" tells CrewAI to use Google embeddings.
• api_key is required to authenticate with Google’s API.

crew = Crew(
    agents=[agent1, agent2],
    tasks=[task1, task2],
    process=Process.sequential,
    memory=True,
    embedder={
        "provider": "google",
        "config": {
            "model": "models/embedding-001",
            "api_key": "YOUR_GOOGLE_API_KEY"
        }
    }
)

crew.kickoff()

Output:

3. Using Hugging Face Embeddings

Hugging Face embeddings give access to a wide range of open-source models. Options like sentence-transformers/all-MiniLM-L6-v2 are lightweight and effective for semantic search and retrieval.

crew = Crew(
    agents=[agent1, agent2],
    tasks=[task1, task2],
    process=Process.sequential,
    memory=True,
    embedder={
        "provider": "huggingface",
        "config": {
            "model": "sentence-transformers/all-MiniLM-L6-v2",
            "api_key": "YOUR_HF_API_KEY"
        }
    }
)

crew.kickoff()

Output:

4. Using Cohere Embeddings

Cohere embeddings are designed for English text and high-performance retrieval tasks. Models like embed-english-v2.0 perform well in semantic similarity and large-scale search.

import cohere

crew = Crew(
    agents=[agent1, agent2],
    tasks=[task1, task2],
    process=Process.sequential,
    memory=True,
    embedder={
        "provider": "cohere",
        "config": {
            "model": "embed-english-v2.0",
            "api_key": "YOUR_COHERE_API_KEY"
        }
    }
)

crew.kickoff()

Output:

Comparison of Embedding Providers in CrewAI

This table outlines the main embedding providers available in CrewAI, showing how they differ in models, strengths and use cases.