Constructing an AI Agent with Llama 4 and AutoGen

Meta’s Llama 4 household of fashions is at present ruling the ever-advancing world of AI. These fashions are revolutionizing how we construct clever techniques with their native multimodal capabilities. When Llama 4 combines with AutoGen, it unlocks the total potential of constructing dynamic, responsive, and sturdy AI Brokers. By leveraging the mixing between Llama 4 and AutoGen, builders can create revolutionary AI brokers that may purpose, collaborate, and adapt effectively. On this article, we’ll discover ways to construct AI brokers with Llama 4 and AutoGen for particular purposes.

Why Ought to We Take into account Utilizing Llama 4?

The Llama 4 mannequin household, together with Scout and Maverick variants, represents a major leap ahead in open-source AI expertise. These fashions provide a number of key benefits:

• Multimodal Intelligence: Llama 4 options native multimodal capabilities that combine various kinds of enter right into a unified structure. This permits extra subtle reasoning throughout totally different media sorts.
• Giant Context Size: It helps as much as 10 million tokens, increasing on Llama 3‘s 128K restrict. It permits dealing with exceptionally lengthy contexts. This makes attainable superior purposes like complete multi-document evaluation, in depth personalization based mostly on person historical past, and navigation of huge codebases.
• Environment friendly Efficiency: Llama 4 employs a Combination of Knowledgeable structure that prompts solely particular parts of the mannequin for every token processed. This strategy makes the fashions extremely environment friendly. Llama 4 Maverick, as an illustration, makes use of simply 17 billion of its complete 400 billion parameters throughout operation. This permits it to run on a single H100 DGX host.
• Superior Efficiency and Capabilities: Benchmark testing reveals Llama 4 Maverick outperforming comparable fashions like GPT-4o and Gemini 2.0 throughout coding, reasoning, multilingual capabilities, and picture understanding.
• Open Supply and Accessible: Meta is making fashions out there for obtain. This encourages open innovation, enabling builders to customise and deploy the expertise throughout various purposes and platforms.

Additionally Learn: DeepSeek V3 vs. Llama 4: Selecting the Proper AI Mannequin for You

Llama 4 Benchmark Efficiency

To know simply how good this mannequin is, right here’s a comparability of Llama 4 towards different prime fashions on numerous normal benchmarks.

 Additionally Learn: Llama 4 vs. GPT-4o: Which is Higher for RAGs?

Constructing an AI Agent Utilizing Llama 4

On this part, I’ll stroll you thru the method of constructing task-specific brokers utilizing Llama 4 and AutoGen. We are going to create a multi-agent system that analyzes shopper necessities for a job, finds freelancers for the actual job based mostly on their expertise and particulars, after which generates customized job proposals for the person to ship out. So let’s start.

Additionally Learn: Palms-on Information to Constructing Multi-Agent Chatbots with AutoGen

Step 0: Setting Up the Atmosphere

Earlier than constructing the agent, we are going to first cowl the required stipulations and arrange the surroundings.

Conditions

Accessing the API

We shall be utilizing the Collectively API right here to entry the Llama 4 mannequin. Create an account on Collectively AI and go to this web page to create your secret key: https://api.collectively.xyz/

Step 1: Organising Libraries and Instruments to Information the AI Brokers

First, we shall be importing all the required libraries and instruments that we’ll want right here.

import os import autogen from IPython.show import show, Markdown

Step 2: Calling the API

To make use of the Llama 4, we now have to load the Collectively API. The code block beneath will assist us load the APIs and configure them to the surroundings.

with open("together_ai_api.txt") as file:    LLAMA_API_KEY = file.learn().strip() os.environ["LLAMA_API_KEY"] = LLAMA_API_KEY

Step 3: Creating Brokers and Defining Duties

Now, let’s create the required brokers and outline their duties, i.e., what they are going to do.

1. Shopper Enter Agent

The Shopper Enter agent acts as the first interface between the human person and the agent system. It collects challenge particulars like shopper necessities, timeline, and finances from the person and passes them to the Scope Architect. It additionally relays follow-up questions and solutions, and alerts termination when the ultimate proposal is accepted.

Anticipated Output:

• Clear transmission of the person’s challenge description and freelancer profile (expertise, expertise, time estimate).
• Ends the session as soon as a passable proposal is delivered, or the person will explicitly finish it.

# Agent 1: Handles Human Enter for Shopper Necessities client_agent = autogen.UserProxyAgent(    identify="Client_Input_Agent",    human_input_mode="ALWAYS",  # asks the human for enter    max_consecutive_auto_reply=1, # Solely reply as soon as    is_termination_msg=lambda x: x.get("content material", "").rstrip().endswith("TERMINATE"),    system_message="""You're the main level of contact for the person.    Your first job is to supply the preliminary challenge particulars obtained from the human person (shopper necessities, product particulars, timeline, finances) to the group chat.    After the Scope Architect asks questions, relay the human person's solutions about their expertise, expertise, instruments, and time estimate again to the chat.    Reply TERMINATE when the ultimate proposal is generated and passable, or if the person needs to cease. In any other case, relay the person's enter.    """, ) 

2. Scope Architect Agent

The Scope Architect Agent is answerable for the preliminary challenge particulars from the Shopper Enter Agent. After that, it asks particular questions to collect the freelancer’s expertise, instruments, previous challenge expertise, and estimated time to finish the work. It doesn’t proceed to proposal technology itself however ensures that each one the required context is collected earlier than handing it over to the following agent.

Anticipated Output:

• Nicely-structured abstract combining each the shopper’s challenge wants and the freelancer’s capabilities.
• Triggers the Price Recommender Agent as soon as all required information is collected and summarized.

# Agent 2: Gathers Consumer's Profile and Estimates scope_architect_agent = autogen.AssistantAgent(    identify="Scope_Architect",    llm_config=llm_config,    human_input_mode="ALWAYS",    max_consecutive_auto_reply=1, # Solely reply as soon as     is_termination_msg=lambda x: x.get("content material", "").rstrip().endswith("TERMINATE"),    system_message="""You're a Scope Architect. Your position is to know the challenge necessities supplied initially after which collect crucial particulars *from the Client_Input_Agent (representing the person/freelancer)*.    1. Look ahead to the preliminary challenge particulars from Client_Input_Agent.    2. After getting the challenge particulars, formulate clear questions for the Client_Input_Agent to ask the human person about their:        - Related previous work/tasks and collaborations.        - Key expertise and instruments relevant to this challenge.        - Their estimated time to finish the outlined work.    3. Do NOT proceed to proposal technology. Look ahead to the Client_Input_Agent to supply the person's solutions.    4. After getting each the shopper necessities AND the person's particulars (expertise, expertise, time estimate), summarize this data clearly for the Price Recommender. Sign that you've all crucial information.    """, ) 

3. Price Recommender Agent

The Price Recommender Agent makes use of the collected data to generate an in depth challenge proposal. It waits for the whole abstract from the Scope Architect. Then analyzes the challenge scope and freelancer particulars to generate knowledgeable proposal doc. This features a customized introduction, a timeline, a number of pricing tiers, and a transparent name to motion.

Anticipated Output:

• Professionally formatted challenge proposal doc with a scope, pricing, and subsequent steps.
• The ultimate output is able to be delivered to the shopper for approval or additional dialogue.

rate_recommender_agent = autogen.AssistantAgent(    identify="Rate_Recommender",    llm_config=llm_config,    max_consecutive_auto_reply=1, # Solely reply as soon as    system_message=f""" You're a Proposal Generator and Price Recommender. Your job is to create a structured challenge proposal. Wait till the Scope_Architect shares a abstract containing BOTH the shopper's challenge necessities AND the person's profile (expertise, expertise, time estimate, previous work if out there). Analyze all obtained information: shopper wants, person experience, estimated time, and any prior fee insights. Generate a well-structured proposal addressed to the shopper, together with the next sections: Customized Introduction: Professionally introduce the person's providers and reference the shopper's firm and challenge. Mission Scope & Timeline: Clearly define the deliverables with estimated timelines based mostly on person enter. Advised Pricing Tiers: Present 1–3 pricing choices (hourly, mounted price, retainer) with justifications based mostly on scope, person expertise, or complexity. Subsequent Steps (CTA): Suggest scheduling a short kickoff name to finalize and make clear particulars. Current ONLY the ultimate formatted proposal. Don't embody further commentary until clarification is requested.""",)

4. Consumer Proxy Agent

This agent acts as an entry level or helper to kick off the interplay. Although it doesn’t play a central position on this stream (based mostly on the code supplied), it could possibly be used to provoke or help with user-facing duties.

user_proxy = autogen.UserProxyAgent(    identify="user_proxy",    max_consecutive_auto_reply=1,    # is_termination_msg=lambda x: x.get("content material", "").rstrip().endswith("TERMINATE"),    llm_config=llm_config,    system_message="""you're an useful assistant and initate the dialog""" )

Step 4: Creating the Group Supervisor

This step units up the central coordinator that manages communication and teamwork between all specialised brokers.

1. Setting Up Group Chat

The Group Chat establishes a structured dialog surroundings for 3 specialised brokers. These are the shopper agent, scope architect agent, and fee recommender agent. It manages dialog stream by means of spherical limits and orderly speaker choice.

Key factors:

• Homes three specialised brokers working towards proposal creation
• 4 rounds most to take care of focus
• “Round_robin” talking sample ensures orderly participation
• Creates a managed surroundings for gathering data

# --- Group Chat Setup --- groupchat = autogen.GroupChat(    brokers=[client_agent, scope_architect_agent, rate_recommender_agent],    messages=[],    max_round=4,    speaker_selection_method="round_robin", )

2. Creating the Group Chat Supervisor

The Group Chat Supervisor orchestrates your complete dialog, guiding interactions by means of a logical development from challenge particulars to proposal technology. Its system message supplies step-by-step directions for agent interactions and defines clear termination situations.

Key factors:

• Directs dialog stream between all brokers
• Hyperlinks to the Group Chat object
• Maintains constant LLM configuration
• Incorporates detailed course of directions
• Terminates upon proposal completion or with the TERMINATE command

supervisor = autogen.GroupChatManager(    groupchat=groupchat,    llm_config=llm_config,    # System message for the supervisor guiding the general stream    system_message="""Handle the dialog stream between the brokers.    1. Begin with the Client_Input_Agent offering challenge particulars.    2. Make sure the Scope_Architect asks the required questions concerning the person's background.    3. Make sure the Client_Input_Agent relays the person's solutions.    4. Make sure the Rate_Recommender waits for all information earlier than producing the ultimate proposal within the specified format.    The dialog finishes when the ultimate proposal is generated or the Client_Input_Agent says TERMINATE.""" )

Step 5: Initiating the Chat

Now that we now have the brokers in place, let’s provoke the collaborative workflow between the brokers. For this, we are going to ship a transparent instruction immediate to the GroupChatManager from the user_proxy agent.

Key factors:

• Triggers the dialog by utilizing user_proxy.initiate_chat(), which begins the group chat and sends the message to the GroupChatManager.
• Delegates management to the supervisor, which then follows the step-by-step stream utilizing the round-robin methodology and its inside system message directions to coordinate the brokers.

# --- Provoke Chat --- print("Beginning the proposal technology course of...") print("Please present the preliminary shopper and challenge particulars when prompted.") initial_prompt_message = """ Begin the method. First, I would like the shopper/challenge particulars from the person (by way of Client_Input_Agent). Then, Scope_Architect ought to ask the person (by way of Client_Input_Agent) about their background. Lastly, Rate_Recommender ought to generate the proposal. """ user_proxy.initiate_chat(    supervisor,    message=initial_prompt_message ) 

Step 6: Formatting the Output

This code will assist us current the output in a markdown(.md) format.

chat_history = supervisor.chat_messages[client_agent] # Or probably simply supervisor.chat_messages if construction differs barely # Discover the final message from the Rate_Recommender agent final_proposal_message = None for msg in reversed(chat_history):    if msg.get("position") == "assistant" and msg.get("identify") == rate_recommender_agent.identify:         if "Customized Introduction:" in msg.get("content material", ""):            final_proposal_message = msg            break if final_proposal_message:    final_proposal_string = final_proposal_message.get("content material", "Proposal content material not discovered.")    strive:        show(Markdown(final_proposal_string))    besides NameError:        print("n(Displaying uncooked Markdown textual content as wealthy output is unavailable)n")        print(final_proposal_string) else:    print("nCould not robotically extract the ultimate proposal from the chat historical past.")    print("You might have to assessment the total chat historical past above.")

Pattern Output

Conclusion

On this article, we constructed a challenge proposal agent utilizing Llama 4 and AutoGen. The agent successfully gathered shopper necessities, structured the proposal, and delivered knowledgeable doc with clear pricing and timeline breakdowns. AutoGen dealt with the dialog stream, whereas Llama 4 ensured pure, context-aware responses all through. This collaboration simplified shopper communication, providing a streamlined resolution for freelancers and consultants to automate proposal technology with minimal guide enter.

Llama 4 enhanced the agent’s efficiency with its improved instruction following, higher context retention, and environment friendly few-shot studying. Its capability to take care of coherence throughout multi-turn dialogues made the proposal technology course of extra clever and responsive. Moreover, the mannequin’s quick inference and low value made it appropriate for real-time purposes. Collectively, Llama 4 and AutoGen allow highly effective agent workflows that increase productiveness and professionalism in client-facing duties.

Often Requested Questions