Deb8flow: Orchestrating Autonomous AI Debates with LangGraph and GPT-4o

I’ve been fascinated by debates—the strategic framing, the sharp retorts, and the rigorously timed comebacks. Debates aren’t simply entertaining; they’re structured battles of concepts, pushed by logic and proof. Lately, I began questioning: might we replicate that dynamic utilizing AI brokers—having them debate one another autonomously, full with real-time fact-checking and moderation? The consequence was Deb8flow, an autonomous AI debating atmosphere powered by LangGraph, OpenAI’s GPT-4o mannequin, and the brand new built-in Net Search function.

In Deb8flow, two brokers—Professional and Con—sq. off on a given matter whereas a Moderator manages turn-taking. A devoted Reality Checker critiques each declare in actual time utilizing GPT-4o’s new shopping capabilities, and a last Decide evaluates the arguments for high quality and coherence. If an agent repeatedly makes factual errors, they’re robotically disqualified—guaranteeing the controversy stays grounded in reality.

This text presents an in-depth take a look at the superior structure and dynamic workflows that energy autonomous AI debates. I’ll stroll you thru how Deb8flow’s modular design leverages LangGraph’s state administration and conditional routing, alongside GPT-4o’s capabilities.

Even in the event you’re new to AI brokers or LangGraph (see assets [1] and [2] for primers), I’ll clarify the important thing ideas clearly. And in the event you’d wish to discover additional, the total challenge is accessible on GitHub: iason-solomos/Deb8flow.

Able to see how AI brokers can debate autonomously in observe?

Let’s dive in.

Excessive-Stage Overview: Autonomous Debates with A number of Brokers

In Deb8flow, we orchestrate a formal debate between two AI brokers – one arguing Professional and one Con – full with a Moderator, a Reality Checker, and a last Decide. The talk unfolds autonomously, with every agent enjoying a task in a structured format.

At its core, Deb8flow is a LangGraph-powered agent system, constructed atop LangChain, utilizing GPT-4o to energy every position—Professional, Con, Decide, and past. We use GPT-4o’s preview mannequin with shopping capabilities to allow real-time fact-checking. In essence, the Professional and Con brokers debate; after every assertion, a fact-checker agent makes use of GPT-4o’s internet search to catch any hallucinations or inaccuracies in that assertion in actual time.​ The talk solely continues as soon as the assertion is verified. The entire course of is coordinated by a LangGraph-defined workflow that ensures correct turn-taking and conditional logic.

The talk workflow goes by way of these levels:

• Subject Technology: A Subject Generator agent produces a nuanced, debatable matter for the session (e.g. “Ought to AI be utilized in classroom schooling?”).
• Opening: The Professional Argument Agent makes a gap assertion in favor of the subject, kicking off the controversy.
• Rebuttal: The Debate Moderator then provides the ground to the Con Argument agent, who rebuts the Professional’s opening assertion.
• Counter: The Moderator provides the ground again to the Professional agent, who counters the Con agent’s factors.
• Closing: The Moderator switches the ground to the Con agent one final time for a closing argument.
• Judgment: Lastly, the Decide agent critiques the total debate historical past and evaluates each side based mostly on argument high quality, readability, and persuasiveness. Probably the most convincing aspect wins.

After each single speech, the Reality Checker agent steps in to confirm the factual accuracy of that assertion​. If a debater’s declare doesn’t maintain up (e.g. cites a unsuitable statistic or “hallucinates” a reality), the workflow triggers a retry: the speaker has to appropriate or modify their assertion. (If both debater accumulates 3 fact-check failures, they’re robotically disqualified for repeatedly spreading inaccuracies, and their opponent wins by default.) This mechanism retains our AI debaters sincere and grounded in actuality!

Conditions and Setup

Earlier than diving into the code, be sure you have the next in place:

• Python 3.12+ put in.
• An OpenAI API key with entry to the GPT-4o mannequin. You possibly can create your personal API key right here: https://platform.openai.com/settings/group/api-keys
• Venture Code: Clone the Deb8flow repository from GitHub (git clone https://github.com/iason-solomos/Deb8flow.git). The repo features a necessities.txt for all required packages. Key dependencies embody LangChain/LangGraph (for constructing the agent graph) and the OpenAI Python consumer.
• Set up Dependencies: In your challenge listing, run: pip set up -r necessities.txt to put in the required libraries.
• Create a .env file within the challenge root to carry your OpenAI API credentials. It needs to be of the shape: OPENAI_API_KEY_GPT4O = "sk-…"
• You may also at any time try the README file: https://github.com/iason-solomos/Deb8flow in the event you merely wish to run the completed app.

As soon as dependencies are put in and the atmosphere variable is about, you ought to be able to run the app. The challenge construction is organized for readability:

Deb8flow/
├── configurations/
│ ├── debate_constants.py
│ └── llm_config.py
├── nodes/
│ ├── base_component.py
│ ├── topic_generator_node.py
│ ├── pro_debater_node.py
│ ├── con_debater_node.py
│ ├── debate_moderator_node.py
│ ├── fact_checker_node.py
│ ├── fact_check_router_node.py
│ └── judge_node.py
├── prompts/
│ ├── topic_generator_prompts.py
│ ├── pro_debater_prompts.py
│ ├── con_debater_prompts.py
│ └── … (prompts for different brokers)
├── assessments/ (comprises unit and complete workflow assessments)
└── debate_workflow.py

A fast tour of this construction:

configurations/ holds fixed definitions and LLM configuration courses.

nodes/ comprises the implementation of every agent or practical node within the debate (every of those is a module defining one agent’s conduct).

prompts/ shops the immediate templates for the language mannequin (so every agent is aware of learn how to immediate GPT-4o for its particular job).

debate_workflow.py ties all the pieces collectively by defining the LangGraph workflow (the graph of nodes and transitions).

debate_state.py defines the shared knowledge construction that the brokers can be utilizing on every run.

assessments/ contains some fundamental assessments and instance runs that will help you confirm all the pieces is working.

Below the Hood: State Administration and Workflow Setup

To coordinate a fancy multi-turn debate, we’d like a shared state and a well-defined movement. We’ll begin by taking a look at how Deb8flow defines the debate state and constants, after which see how the LangGraph workflow is constructed.

Defining the Debate State Schema (debate_state.py)

Deb8flow makes use of a shared state (https://langchain-ai.github.io/langgraph/ideas/low_level/#state ) within the type of a Python TypedDict that every one brokers can learn from and replace. This state tracks the controversy’s progress and context – issues like the subject, the historical past of messages, whose flip it’s, and many others. By centralizing this info, every agent node could make choices based mostly on the present state of the controversy.

Hyperlink: debate_state.py

from typing import TypedDict, Record, Dict, Literal


DebateStage = Literal["opening", "rebuttal", "counter", "final_argument"]

class DebateMessage(TypedDict):
    speaker: str  # e.g. professional or con
    content material: str  # The message every speaker produced
    validated: bool  # Whether or not the FactChecker okay’d this message
    stage: DebateStage # The stage of the controversy when this message was produced

class DebateState(TypedDict):
    debate_topic: str
    positions: Dict[str, str]
    messages: Record[DebateMessage]
    opening_statement_pro_agent: str
    stage: str  # "opening", "rebuttal", "counter", "final_argument"
    speaker: str  # "professional" or "con"
    times_pro_fact_checked: int # The variety of occasions the professional agent has been fact-checked. If it reaches 3, the professional agent is disqualified.
    times_con_fact_checked: int # The variety of occasions the con agent has been fact-checked. If it reaches 3, the con agent is disqualified.

Key fields that we have to have within the DebateState embody:

• debate_topic (str): The subject being debated.
• messages (Record[DebateMessage]): A listing of all messages exchanged to date. Every message is a dictionary with fields for speaker (e.g. "professional" or "con" or "fact_checker"), the message content material (textual content), a validated flag (whether or not it handed fact-check), and the stage of the controversy when it was produced.
• stage (str): The present debate stage (one among "opening", "rebuttal", "counter", "final_argument").
• speaker (str): Whose flip it’s at the moment ("professional" or "con").
• times_pro_fact_checked / times_con_fact_checked (int): Counters for what number of occasions both sides has been caught with a false declare. (In our guidelines, if a debater fails fact-check 3 occasions, they may very well be disqualified or robotically lose.)
• positions (Dict[str, str]): (Elective) A mapping of every aspect’s normal stance (e.g., "professional": "In favor of the subject").

By structuring the controversy’s state, brokers discover it straightforward to entry the dialog historical past or verify the present stage, and the management logic can replace the state between turns. The state is basically the reminiscence of the controversy.

Constants and Configuration

To keep away from “magic strings” scattered within the code, we outline some constants in debate_constants.py. For instance, constants for stage names (STAGE_OPENING = "opening", and many others.), speaker identifiers (SPEAKER_PRO = "professional", SPEAKER_CON = "con", and many others.), and node names (NODE_PRO_DEBATER = "pro_debater_node", and many others.). These make the code simpler to take care of and skim.

debate_constants.py:

# Stage names
STAGE_OPENING = "opening"
STAGE_REBUTTAL = "rebuttal"
STAGE_COUNTER = "counter"
STAGE_FINAL_ARGUMENT = "final_argument"
STAGE_END = "finish"

# Audio system
SPEAKER_PRO = "professional"
SPEAKER_CON = "con"
SPEAKER_JUDGE = "decide"

# Node names
NODE_PRO_DEBATER = "pro_debater_node"
NODE_CON_DEBATER = "con_debater_node"
NODE_DEBATE_MODERATOR = "debate_moderator_node"
NODE_JUDGE = "judge_node"

We additionally arrange LLM configuration in llm_config.py. Right here, we outline courses for OpenAI or Azure OpenAI configs after which create a dictionary llm_config_map mapping mannequin names to their config. As an example, we map "gpt-4o" to an OpenAILLMConfig that holds the mannequin title and API key. This manner, every time we have to initialize a GPT-4o agent, we will simply do llm_config_map["gpt-4o"] to get the precise config. All our primary brokers (debaters, matter generator, decide) use this identical GPT-4o configuration.

import os
from dataclasses import dataclass
from typing import Union

@dataclass
class OpenAILLMConfig:
    """
    A knowledge class to retailer configuration particulars for OpenAI fashions.

    Attributes:
        model_name (str): The title of the OpenAI mannequin to make use of.
        openai_api_key (str): The API key for authenticating with the OpenAI service.
    """
    model_name: str
    openai_api_key: str


llm_config_map = {
    "gpt-4o": OpenAILLMConfig(
        model_name="gpt-4o",
        openai_api_key=os.getenv("OPENAI_API_KEY_GPT4O"),
    )
}

Constructing the LangGraph Workflow (debate_workflow.py)

With state and configs in place, we assemble the debate workflow graph. LangGraph’s StateGraph is the spine that connects all our agent nodes within the order they need to execute. Right here’s how we set it up:

class DebateWorkflow:

    def _initialize_workflow(self) -> StateGraph:
        workflow = StateGraph(DebateState)
        # Nodes
        workflow.add_node("generate_topic_node", GenerateTopicNode(llm_config_map["gpt-4o"]))
        workflow.add_node("pro_debater_node", ProDebaterNode(llm_config_map["gpt-4o"]))
        workflow.add_node("con_debater_node", ConDebaterNode(llm_config_map["gpt-4o"]))
        workflow.add_node("fact_check_node", FactCheckNode())
        workflow.add_node("fact_check_router_node", FactCheckRouterNode())
        workflow.add_node("debate_moderator_node", DebateModeratorNode())
        workflow.add_node("judge_node", JudgeNode(llm_config_map["gpt-4o"]))

        # Entry level
        workflow.set_entry_point("generate_topic_node")

        # Movement
        workflow.add_edge("generate_topic_node", "pro_debater_node")
        workflow.add_edge("pro_debater_node", "fact_check_node")
        workflow.add_edge("con_debater_node", "fact_check_node")
        workflow.add_edge("fact_check_node", "fact_check_router_node")
        workflow.add_edge("judge_node", END)
        return workflow



    async def run(self):
        workflow = self._initialize_workflow()
        graph = workflow.compile()
        # graph.get_graph().draw_mermaid_png(output_file_path="workflow_graph.png")
        initial_state = {
            "matter": "",
            "positions": {}
        }
        final_state = await graph.ainvoke(initial_state, config={"recursion_limit": 50})
        return final_state

Let’s break down what’s taking place:

• We initialize a brand new StateGraph with our DebateState kind because the state schema.
• We add every node (agent) to the graph with a reputation. For nodes that want an LLM, we move within the GPT-4o config. For instance, "pro_debater_node" is added as ProDebaterNode(llm_config_map["gpt-4o"]), that means the Professional debater agent will use GPT-4o as its underlying mannequin.
• We set the entry level of the graph to "generate_topic_node". This implies step one of the workflow is to generate a debate matter.
• Then we add directed edges to attach nodes. The perimeters above encode the first sequence: matter -> professional’s flip -> fact-check -> (then a routing resolution) -> … finally -> decide -> END. We don’t join the Moderator or Reality Examine Router with static edges, since these nodes use dynamic instructions to redirect the movement. The ultimate edge connects the decide to an END marker to terminate the graph.

When the workflow runs, management will move alongside these edges so as, however every time we hit a router or moderator node, that node will output a command telling the graph which node to go to subsequent (overriding the default edge). That is how we create conditional loops: the fact_check_router_node may ship us again to a debater node for a retry, as a substitute of following a straight line. LangGraph helps this by permitting nodes to return a particular Command object with goto directions.

In abstract, at a excessive degree we’ve outlined an agentic workflow: a graph of autonomous brokers the place management can department and loop based mostly on the brokers’ outputs. Now, let’s discover what every of those agent nodes truly does.

Agent Nodes Breakdown

Every stage or position within the debate is encapsulated in a node (agent). In LangGraph, nodes are sometimes easy features, however I needed a extra object-oriented strategy for readability and reusability. So in Deb8flow, each node is a class with a __call__ technique. All the primary agent courses inherit from a standard BaseComponent for shared performance. This design makes the system modular: we will simply swap out or lengthen brokers by modifying their class definitions, and every agent class is accountable for its piece of the workflow.

Let’s undergo the important thing brokers one after the other.

BaseComponent – A Reusable Agent Base Class

Most of our agent nodes (just like the debaters and decide) share frequent wants: they use an LLM to generate output, they could must retry on errors, and they need to observe token utilization. The BaseComponent class (outlined in nodes/base_component.py) offers these frequent options so we don’t repeat code.

class BaseComponent:
    """
    A foundational class for managing LLM-based workflows with token monitoring.
    Can deal with each Azure OpenAI (AzureChatOpenAI) and OpenAI (ChatOpenAI).
    """

    def __init__(
        self,
        llm_config: Elective[LLMConfig] = None,
        temperature: float = 0.0,
        max_retries: int = 5,
    ):
        """
        Initializes the BaseComponent with non-obligatory LLM configuration and temperature.

        Args:
            llm_config (Elective[LLMConfig]): Configuration for both Azure or OpenAI.
            temperature (float): Controls the randomness of LLM outputs. Defaults to 0.0.
            max_retries (int): What number of occasions to retry on 429 errors.
        """
        logger = logging.getLogger(self.__class__.__name__)
        tracer = hint.get_tracer(__name__, tracer_provider=get_tracer_provider())

        self.logger = logger
        self.tracer = tracer
        self.llm: Elective[ChatOpenAI] = None
        self.output_parser: Elective[StrOutputParser] = None
        self.state: Elective[DebateState] = None
        self.prompt_template: Elective[ChatPromptTemplate] = None
        self.chain: Elective[RunnableSequence] = None
        self.paperwork: Elective[List] = None
        self.prompt_tokens = 0
        self.completion_tokens = 0
        self.max_retries = max_retries

        if llm_config isn't None:
            self.llm = self._init_llm(llm_config, temperature)
            self.output_parser = StrOutputParser()

    def _init_llm(self, config: LLMConfig, temperature: float):
        """
        Initializes an LLM occasion for both Azure OpenAI or customary OpenAI.
        """
        if isinstance(config, AzureOpenAILLMConfig):
            # If it is Azure, use the AzureChatOpenAI class
            return AzureChatOpenAI(
                deployment_name=config.deployment_name,
                azure_endpoint=config.azure_endpoint,
                openai_api_version=config.openai_api_version,
                openai_api_key=config.openai_api_key,
                temperature=temperature,
            )
        elif isinstance(config, OpenAILLMConfig):
            # If it is customary OpenAI, use the ChatOpenAI class
            return ChatOpenAI(
                model_name=config.model_name,
                openai_api_key=config.openai_api_key,
                temperature=temperature,
            )
        else:
            increase ValueError("Unsupported LLMConfig kind.")

    def validate_initialization(self) -> None:
        """
        Ensures we now have an LLM and an output parser.
        """
        if not self.llm:
            increase ValueError("LLM isn't initialized. Guarantee `llm_config` is supplied.")
        if not self.output_parser:
            increase ValueError("Output parser isn't initialized.")

    def execute_chain(self, inputs: Any) -> Any:
        """
        Executes the LLM chain, tracks token utilization, and retries on 429 errors.
        """
        if not self.chain:
            increase ValueError("No chain is initialized for execution.")

        retry_wait = 1  # Preliminary wait time in seconds

        for try in vary(self.max_retries):
            attempt:
                with get_openai_callback() as cb:
                    consequence = self.chain.invoke(inputs)
                    self.logger.data("Immediate Token utilization: %s", cb.prompt_tokens)
                    self.logger.data("Completion Token utilization: %s", cb.completion_tokens)
                    self.prompt_tokens = cb.prompt_tokens
                    self.completion_tokens = cb.completion_tokens

                return consequence

            besides Exception as e:
                # If the error mentions 429, do exponential backoff and retry
                if "429" in str(e):
                    self.logger.warning(
                        f"Charge restrict reached. Retrying in {retry_wait} seconds... "
                        f"(Try {try + 1}/{self.max_retries})"
                    )
                    time.sleep(retry_wait)
                    retry_wait *= 2
                else:
                    self.logger.error(f"Sudden error: {str(e)}")
                    increase e

        increase Exception("API request failed after most variety of retries")

    def create_chain(
        self, system_template: str, human_template: str
    ) -> RunnableSequence:
        """
        Creates a sequence for unstructured outputs.
        """
        self.validate_initialization()
        self.prompt_template = ChatPromptTemplate.from_messages(
            [
                ("system", system_template),
                ("human", human_template),
            ]
        )
        self.chain = self.prompt_template | self.llm | self.output_parser
        return self.chain

    def create_structured_output_chain(
        self, system_template: str, human_template: str, output_model: Kind[BaseModel]
    ) -> RunnableSequence:
        """
        Creates a sequence that yields structured outputs (parsed right into a Pydantic mannequin).
        """
        self.validate_initialization()
        self.prompt_template = ChatPromptTemplate.from_messages(
            [
                ("system", system_template),
                ("human", human_template),
            ]
        )
        self.chain = self.prompt_template | self.llm.with_structured_output(output_model)
        return self.chain

    def build_return_with_tokens(self, node_specific_data: dict) -> dict:
        """
        Comfort technique so as to add token utilization data into the return values.
        """
        return {
            **node_specific_data,
            "prompt_tokens": self.prompt_tokens,
            "completion_tokens": self.completion_tokens,
        }

    def __call__(self, state: DebateState) -> None:
        """
        Updates the node's native copy of the state.
        """
        self.state = state
        for key, worth in state.objects():
            setattr(self, key, worth)

Key options of BaseComponent:

• It shops an LLM consumer (e.g. an OpenAI ChatOpenAI occasion) initialized with a given mannequin and API key, in addition to an output parser.
• It offers a way create_chain(system_template, human_template) which units up a LangChain immediate chain (a RunnableSequence) combining a system immediate and a human immediate. This chain is what truly generates outputs when run.
• It has an execute_chain(inputs) technique that invokes the chain and contains logic to retry if the OpenAI API returns a rate-limit error (HTTP 429). That is achieved with exponential backoff as much as a max_retries depend.
• It retains observe of token utilization (immediate tokens and completion tokens) for logging or evaluation.
• The __call__ technique of BaseComponent (which every subclass will name by way of tremendous().__call__(state)) can carry out any setup wanted earlier than the node’s primary logic runs (like guaranteeing the LLM is initialized).

By constructing on BaseComponent, every agent class can deal with its distinctive logic (like what immediate to make use of and learn how to deal with the state), whereas inheriting the heavy lifting of interacting with GPT-4o reliably.

Subject Generator Agent (GenerateTopicNode)

The Subject Generator (topic_generator_node.py) is the primary agent within the graph. Its job is to give you a debatable matter for the session. We give it a immediate that instructs it to output a nuanced matter that would moderately have a professional and con aspect.

This agent inherits from BaseComponent and makes use of a immediate chain (system + human immediate) to generate one merchandise of textual content – the controversy matter. When known as, it executes the chain (with no particular enter, simply utilizing the immediate) and will get again a topic_text. It then updates the state with:

• debate_topic: the generated matter (stripped of any further whitespace),
• positions: a dictionary assigning the professional and con stances (by default we use "In favor of the subject" and "Towards the subject"),
• stage: set to "opening",
• speaker: set to "professional" (so the Professional aspect will converse first).

In code, the return may appear to be:

return {
    "debate_topic": debate_topic,
    "positions": positions,
    "stage": "opening",
    "speaker": first_speaker  # "professional"
}

Listed below are the prompts for the subject generator:

SYSTEM_PROMPT = """
You're a brainstorming AI that means debate matters.
You'll present a single, fascinating or well timed matter that may have two opposing views.
"""

HUMAN_PROMPT = """
Please recommend one debate matter for 2 AI brokers to debate.
For instance, it may very well be about know-how, politics, philosophy, or any fascinating area.
Simply present the subject in a concise sentence.
"""

Then we move these prompts within the constructor of the category itself.

class GenerateTopicNode(BaseComponent):
    def __init__(self, llm_config, temperature: float = 0.7):
        tremendous().__init__(llm_config, temperature)
        # Create the immediate chain.
        self.chain: RunnableSequence = self.create_chain(
            system_template=SYSTEM_PROMPT,
            human_template=HUMAN_PROMPT
        )

    def __call__(self, state: DebateState) -> Dict[str, str]:
        """
        Generates a debate matter and assigns positions to the 2 debaters.
        """
        tremendous().__call__(state)

        topic_text = self.execute_chain({})

        # Retailer the subject and assign stances within the DebateState
        debate_topic = topic_text.strip()
        positions = {
            "professional": "In favor of the subject",
            "con": "Towards the subject"
        }

        
        first_speaker = "professional"
        self.logger.data("Welcome to our debate panel! In the present day's debate matter is: %s", debate_topic)
        return {
            "debate_topic": debate_topic,
            "positions": positions,
            "stage": "opening",
            "speaker": first_speaker
        }

It’s a sample we’ll repeat for all courses aside from these not utilizing LLMs and the actual fact checker.

Now we will implement the two stars of the present, the Professional and Con argument brokers!

Debater Brokers (Professional and Con)

Hyperlink: pro_debater_node.py

The 2 debater brokers are very related in construction, however every makes use of completely different immediate templates tailor-made to their position (professional vs con) and the stage of the controversy.

The Professional debater, for instance, has to deal with an opening assertion and a counter-argument (countering the Con’s rebuttal). We additionally want logic for retries in case an announcement fails fact-check. In code, the ProDebater class units up a number of immediate chains:

• opening_chain and an opening_retry_chain (utilizing barely completely different human prompts – the retry immediate may instruct it to attempt once more with out repeating any factually doubtful claims).
• counter_chain and counter_retry_chain for the counter-argument stage.

class ProDebaterNode(BaseComponent):
    def __init__(self, llm_config, temperature: float = 0.7):
        tremendous().__init__(llm_config, temperature)
        self.opening_chain = self.create_chain(SYSTEM_PROMPT, OPENING_HUMAN_PROMPT)
        self.opening_retry_chain = self.create_chain(SYSTEM_PROMPT, OPENING_RETRY_HUMAN_PROMPT)
        self.counter_chain = self.create_chain(SYSTEM_PROMPT, COUNTER_HUMAN_PROMPT)
        self.counter_retry_chain = self.create_chain(SYSTEM_PROMPT, COUNTER_RETRY_HUMAN_PROMPT)

    def __call__(self, state: DebateState) -> Dict[str, Any]:
        tremendous().__call__(state)

        debate_topic = state.get("debate_topic")
        messages = state.get("messages", [])
        stage = state.get("stage")
        speaker = state.get("speaker")

        # Examine if retrying (final message was by professional and never validated)
        last_msg = messages[-1] if messages else None
        retrying = last_msg and last_msg["speaker"] == SPEAKER_PRO and never last_msg["validated"]

        if stage == STAGE_OPENING and speaker == SPEAKER_PRO:
            chain = self.opening_retry_chain if retrying else self.opening_chain # choose which chain we're triggering: the conventional one or the fact-cehcked one
            consequence = chain.invoke({
                "debate_topic": debate_topic
            })
        elif stage == STAGE_COUNTER and speaker == SPEAKER_PRO:
            opponent_msg = self._get_last_message_by(SPEAKER_CON, messages)
            debate_history = get_debate_history(messages)
            chain = self.counter_retry_chain if retrying else self.counter_chain
            consequence = chain.invoke({
                "debate_topic": debate_topic,
                "opponent_statement": opponent_msg,
                "debate_history": debate_history
            })
        else:
            increase ValueError(f"Unknown flip for ProDebater: stage={stage}, speaker={speaker}")
        new_message = create_debate_message(speaker=SPEAKER_PRO, content material=consequence, stage=stage)
        self.logger.data("Speaker: %s, Stage: %s, Retry: %snMessage:npercents", speaker, stage, retrying, consequence)
        return {
            "messages": messages + [new_message]
        }

    def _get_last_message_by(self, speaker_prefix, messages):
        for m in reversed(messages):
            if m.get("speaker") == speaker_prefix:
                return m["content"]
        return ""

When the ProDebater’s __call__ runs, it seems on the present stage and speaker within the state to determine what to do:

• If it’s the opening stage and the speaker is “professional”, it makes use of the opening_chain to generate a gap argument. If the final message from Professional was marked invalid (not validated), it is aware of it is a retry, so it will use the opening_retry_chain as a substitute.
• If it’s the counter stage and speaker is “professional”, it generates a counter-argument to regardless of the opponent (Con) simply mentioned. It should fetch the final message by the Con from the messages historical past, and feed that into the immediate (in order that the Professional can straight counter it). Once more, if the final Professional message was invalid, it will swap to the retry chain.

After producing its argument, the Debater agent creates a brand new message entry (with speaker="professional", the content material textual content, validated=False initially, and the stage) and appends it to the state’s message record. That turns into the output of the node (LangGraph will merge this partial state replace into the worldwide state).

The Con Debater agent mirrors this logic for its levels:

It equally appends its message to the state.

It has a rebuttal and closing argument (last argument) stage, every with a traditional and a retry chain.

It checks if it’s the rebuttal stage (speaker “con”) or last argument stage (speaker “con”) and invokes the suitable chain, presumably utilizing the final Professional message for context when rebutting.

con_debater_node.py

By utilizing class-based implementation, our debaters’ code is simpler to take care of. We are able to clearly separate what the Professional does vs what the Con does, even when they share construction. Additionally, by encapsulating immediate chains inside the category, every debater can handle a number of doable outputs (common vs retry) cleanly.

Immediate design: The precise prompts (in prompts/pro_debater_prompts.py and con_debater_prompts.py) information the GPT-4o mannequin to tackle a persona (“You’re a debater arguing for/towards the subject…”) and produce the argument. Additionally they instruct the mannequin to maintain statements factual and logical. If a reality verify fails, the retry immediate could say one thing like: “Your earlier assertion had an unverified declare. Revise your argument to be factually appropriate whereas sustaining your place.” – encouraging the mannequin to appropriate itself.

With this, our AI debaters can interact in a multi-turn duel, and even get better from factual missteps.

Reality Checker Agent (FactCheckNode)

After every debater speaks, the Reality Checker agent swoops in to confirm their claims. This agent is applied in fact_checker_node.py, and curiously, it makes use of the GPT-4o mannequin’s shopping capacity fairly than our personal customized prompts. Primarily, we delegate the fact-checking to OpenAI’s GPT-4 with internet search.

How does this work? The OpenAI Python consumer for GPT-4 (with shopping) permits us to ship a person message and get a structured response. In FactCheckNode.__call__, we do one thing like:

completion = self.consumer.beta.chat.completions.parse(
            mannequin="gpt-4o-search-preview",
            web_search_options={},
            messages=[{
                "role": "user",
                "content": (
                        f"Consider the following statement from a debate. "
                        f"If the statement contains numbers, or figures from studies, fact-check it online.nn"
                        f"Statement:n"{claim}"nn"
                        f"Reply clearly whether any numbers or studies might be inaccurate or hallucinated, and why."
                        f"n"
                        f"If the statement doesn't contain references to studies or numbers cited, don't go online to fact-check, and just consider it successfully fact-checked, with a 'yes' score.nn"
                )
            }],
            response_format=FactCheck
        )

If the result’s “sure” (that means the declare appears truthful or no less than not factually unsuitable), the Reality Checker will mark the final message’s validated area as True within the state, and output {"validated": True} with no additional adjustments. This indicators that the controversy can proceed usually.

If the result’s “no” (that means it discovered the declare to be incorrect or doubtful), the Reality Checker will append a brand new message to the state with speaker="fact_checker" describing the discovering (or we might merely mark it, however offering a quick be aware like “(Reality Checker: The statistic cited couldn’t be verified.)” may be helpful). It should additionally set validated: False and increment a counter for whichever aspect made the declare. The output state from this node contains validated: False and an up to date times_pro_fact_checked or times_con_fact_checked depend.

We additionally use a Pydantic BaseModel to manage the output of the LLM:

class FactCheck(BaseModel):
    """
    Pydantic mannequin for the actual fact checking the claims made by debaters.

    Attributes:
        binary_score (str): 'sure' if the declare is verifiable and truthful, 'no' in any other case.
    """

    binary_score: str = Discipline(
        description="Signifies if the declare is verifiable and truthful. 'sure' or 'no'."
    )
    justification: str = Discipline(
        description="Clarification of the reasoning behind the rating."
    )

Debate Moderator Agent (DebateModeratorNode)

The Debate Moderator is the conductor of the controversy. As an alternative of manufacturing prolonged textual content, this agent’s job is to handle turn-taking and stage development. Within the workflow, after an announcement is validated by the Reality Checker, management passes to the Moderator node. The Moderator then points a Command that updates the state for the subsequent flip and directs the movement to the suitable subsequent agent.

The logic in DebateModeratorNode.__call__ (see nodes/debate_moderator_node.py) goes roughly like this:

if stage == STAGE_OPENING and speaker == SPEAKER_PRO:
            return Command(
                replace={"stage": STAGE_REBUTTAL, "speaker": SPEAKER_CON},
                goto=NODE_CON_DEBATER
            )
        elif stage == STAGE_REBUTTAL and speaker == SPEAKER_CON:
            return Command(
                replace={"stage": STAGE_COUNTER, "speaker": SPEAKER_PRO},
                goto=NODE_PRO_DEBATER
            )
        elif stage == STAGE_COUNTER and speaker == SPEAKER_PRO:
            return Command(
                replace={"stage": STAGE_FINAL_ARGUMENT, "speaker": SPEAKER_CON},
                goto=NODE_CON_DEBATER
            )
        elif stage == STAGE_FINAL_ARGUMENT and speaker == SPEAKER_CON:
            return Command(
                replace={},
                goto=NODE_JUDGE
            )

        increase ValueError(f"Sudden stage/speaker combo: stage={stage}, speaker={speaker}")

Every conditional corresponds to some extent within the debate the place a flip simply ended, and units up the subsequent flip. For instance, after the opening (Professional simply spoke), it units stage to rebuttal, switches speaker to Con, and directs the workflow to the Con debater node​. After the final_argument (Con’s closing), it directs to the Decide with no additional replace (the controversy stage successfully ends).

Reality Examine Router (FactCheckRouterNode)

That is one other management node (just like the Moderator) that introduces conditional logic. The Reality Examine Router sits proper after the Reality Checker agent within the movement. Its objective is to department the workflow relying on the fact-check consequence.

In nodes/fact_check_router_node.py, the logic is:

if pro_fact_checks >= 3 or con_fact_checks >= 3:
            disqualified = SPEAKER_PRO if pro_fact_checks >= 3 else SPEAKER_CON
            winner = SPEAKER_CON if disqualified == SPEAKER_PRO else SPEAKER_PRO

            verdict_msg = {
                "speaker": "moderator",
                "content material": (
                    f"Debate ended early because of extreme factual inaccuracies.nn"
                    f"DISQUALIFIED: {disqualified.higher()} (exceeded reality verify restrict)n"
                    f"WINNER: {winner.higher()}"
                ),
                "validated": True,
                "stage": "verdict"
            }
            return Command(
                replace={"messages": messages + [verdict_msg]},
                goto=END
            )
        if last_message.get("validated"):
            return Command(goto=NODE_DEBATE_MODERATOR)
        elif speaker == SPEAKER_PRO:
            return Command(goto=NODE_PRO_DEBATER)
        elif speaker == SPEAKER_CON:
            return Command(goto=NODE_CON_DEBATER)
        increase ValueError("Unable to find out routing in FactCheckRouterNode.")

First, the Reality Examine Router checks if both aspect’s fact-check depend has reached 3. In that case, it creates a Moderator-style message saying an early finish: the offending aspect is disqualified and the opposite aspect is the winner​. It appends this verdict to the messages and returns a Command that jumps to END, successfully terminating the controversy with out going to the Decide (as a result of we already know the result).

If we’re not ending the controversy early, it then seems on the Reality Checker’s consequence for the final message (which is saved as validated on that message). If validated is True, we go to the controversy moderator: Command(goto=debate_moderator_node).

Else if the assertion fails fact-check, the workflow goes again to the debater to supply a revised assertion (with the state counters up to date to mirror the failure). This loop can occur a number of occasions if wanted (as much as the disqualification restrict).

This dynamic management is the guts of Deb8flow’s “agentic” nature – the power to adapt the trail of execution based mostly on the content material of the brokers’ outputs. It showcases LangGraph’s energy: combining management movement with state. We’re primarily encoding debate guidelines (like permitting retries for false claims, or ending the controversy if somebody cheats too typically) straight into the workflow graph.

Decide Agent (JudgeNode)

Final however not least, the Decide agent delivers the ultimate verdict based mostly on rhetorical ability, readability, construction, and total persuasiveness. Its system immediate and human immediate make this express:

• System Immediate: “You’re an neutral debate decide AI. … Consider which debater introduced their case extra clearly, persuasively, and logically. You should deal with communication abilities, construction of argument, rhetorical energy, and total coherence.”
• Human Immediate: “Right here is the total debate transcript. Please analyze the efficiency of each debaters—PRO and CON. Consider rhetorical efficiency—readability, construction, persuasion, and relevance—and determine who introduced their case extra successfully.”

When the Decide node runs, it receives the complete debate transcript (all validated messages) alongside the unique matter. It then makes use of GPT-4o to look at how both sides framed their arguments, dealt with counterpoints, and supported (or didn’t help) claims with examples or logic. Crucially, the Decide is forbidden to judge which place is objectively appropriate (or who it thinks could be appropriate)—solely who argued extra persuasively.

Beneath is an instance last verdict from a Deb8flow run on the subject:
“Ought to governments implement a common fundamental revenue in response to rising automation within the workforce?”

WINNER: PRO

REASON: The PRO debater introduced a extra compelling and rhetorically efficient case for common fundamental revenue. Their arguments have been well-structured, starting with a transparent assertion of the problem and the need of UBI in response to automation. They successfully addressed potential counterarguments by highlighting the unprecedented velocity and scope of present technological adjustments, which distinguishes the present state of affairs from previous technological shifts. The PRO additionally supplied empirical proof from UBI pilot applications to counter the CON's claims about work disincentives and financial inefficiencies, reinforcing their argument with real-world examples.

In distinction, the CON debater, whereas presenting legitimate issues about UBI, relied closely on historic analogies and assumptions about workforce adaptability with out adequately addressing the distinctive challenges posed by trendy automation. Their arguments concerning the fiscal burden and potential inefficiencies of UBI have been much less supported by particular proof in comparison with the PRO's rebuttals.

Total, the PRO's arguments have been extra coherent, persuasive, and backed by empirical proof, making their case extra convincing to a impartial observer.

Langsmith Tracing

All through Deb8flow’s improvement, I relied on LangSmith (LangChain’s tracing and observability toolkit) to make sure the complete debate pipeline was behaving accurately. As a result of we now have a number of brokers passing management between themselves, it’s straightforward for surprising loops or misrouted states to happen. LangSmith offers a handy method to:

• Visualize Execution Movement: You possibly can see every agent’s immediate, the tokens consumed (so you too can observe prices), and any intermediate states. This makes it a lot easier to substantiate that, say, the Con Debater is correctly referencing the Professional Debater’s final message, or that the Reality Checker is precisely receiving the declare to confirm.
• Debug State Updates: If the Moderator or Reality Examine Router is sending the movement to the unsuitable node, the hint will spotlight that mismatch. You possibly can hint which agent was invoked at every step and why, serving to you notice stage or speaker misalignments early.
• Observe Immediate and Completion Tokens: With a number of GPT-4o calls, it’s helpful to see what number of tokens every stage is utilizing, which LangSmith logs robotically in the event you allow tracing.

Integrating LangSmith is unexpectedly straightforward. You’ll simply want to offer these 3 keys in your .env file: LANGCHAIN_API_KEY

LANGCHAIN_TRACING_V2

LANGCHAIN_PROJECT

Then you’ll be able to open the LangSmith UI to see a structured hint of every run. This tremendously reduces the guesswork concerned in debugging multi-agent techniques and is, in my expertise, important for extra advanced AI orchestration like ours. Instance of a single run:

Reflections and Subsequent Steps

Constructing Deb8flow was an eye-opening train in orchestrating autonomous agent workflows. We didn’t simply chain a single mannequin name – we created a whole debate simulation with AI brokers, every with a selected position, and allowed them to work together in accordance with a algorithm. LangGraph supplied a transparent framework to outline how knowledge and management flows between brokers, making the advanced sequence manageable in code. By utilizing class-based brokers and a shared state, we maintained modularity and readability, which is able to repay for any software program engineering challenge in the long term.

An thrilling facet of this challenge was seeing emergent conduct. Although every agent follows a script (a immediate), the unscripted mixture – a debater making an attempt to deceive, a fact-checker catching it, the debater rephrasing – felt surprisingly reasonable! It’s a small step towards extra Agentic Ai techniques that may carry out non-trivial multi-step duties with oversight on one another.

There’s loads of concepts for enchancment:

• Consumer Interplay: Presently it’s absolutely autonomous, however one might add a mode the place a human offers the subject and even takes the position of 1 aspect towards an AI opponent.
• We are able to swap the order by which the Debaters speak.
• We are able to change the prompts, and thus to an excellent diploma the conduct of the brokers, and experiment with completely different prompts.
• Make the debaters additionally carry out internet search earlier than producing their statements, thus offering them with the most recent info.

The broader implication of Deb8flow is the way it showcases a sample for composable AI brokers. By defining clear boundaries and interactions (similar to microservices in software program), we will have advanced AI-driven processes that stay interpretable and controllable. Every agent is sort of a cog in a machine, and LangGraph is the gear system making them work in unison.

I discovered this challenge energizing, and I hope it conjures up you to discover multi-agent workflows. Whether or not it’s debating, collaborating on writing, or fixing issues from completely different professional angles, the mixture of GPT, instruments, and structured agentic workflows opens up a brand new world of potentialities for AI improvement. Blissful hacking!

References

[1] D. Bouchard, “From Fundamentals to Superior: Exploring LangGraph,” Medium, Nov. 22, 2023. [Online]. Accessible: https://medium.com/data-science/from-basics-to-advanced-exploring-langgraph-e8c1cf4db787. [Accessed: Apr. 1, 2025].

[2] A. W. T. Ng, “Constructing a Analysis Agent that Can Write to Google Docs: Half 1,” In the direction of Information Science, Jan. 11, 2024. [Online]. Accessible: https://towardsdatascience.com/building-a-research-agent-that-can-write-to-google-docs-part-1-4b49ea05a292/. [Accessed: Apr. 1, 2025].