​
🧠 Conversation with Memory

A conversation with memory is more than just talk—it’s connection.

This guide demonstrates how to create an AI conversation pipeline using the superagentx framework, with built-in support for memory and conversation context through a unique conversation_id.

The core idea is to enable an AI agent to remember the flow of a conversation using a memory component tied to a specific session.

​
1. Define the AI Handler Pipeline

The function get_test_ai_handler_pipe initializes the AI pipeline with:

  • An LLM client (using OpenAI)
  • A memory module to retain context
  • A custom handler to manage LLM interactions
  • An agent defined with a goal and role
  • A pipeline (AgentXPipe) for integration into CLI/WebSocket/API interfaces
import uuid
from superagentx.agent import Agent
from superagentx.agentxpipe import AgentXPipe
from superagentx.engine import Engine
from superagentx.llm import LLMClient
from superagentx.memory import Memory
from superagentx.prompt import PromptTemplate

from df_agentic_app.handlers.ai import AIHandler

async def get_test_ai_handler_pipe() -> AgentXPipe:
    # LLM Configuration
    llm_config = {
        'llm_type': 'openai'
    }
    llm_client = LLMClient(llm_config=llm_config)

    # Enable Memory
    memory = Memory(memory_config={"llm_client": llm_client})

    ai_handler = AIHandler(
        llm=llm_client,
        memory=memory
    )

    # Prompt Template
    prompt_template = PromptTemplate()

    # Create Engine
    ai_engine = Engine(
        handler=ai_handler,
        llm=llm_client,
        prompt_template=prompt_template
    )

    # Create Agents
    ai_agent = Agent(
        name='AI agent',
        goal="Get me the best results",
        role="You are the best at answering",
        llm=llm_client,
        prompt_template=prompt_template,
        engines=[ai_engine]
    )

    # Expose via pipe for interaction
    pipe = AgentXPipe(
        agents=[ai_agent],
        memory=memory
    )

    return pipe

​
2. Run a Conversation with Context

This example initializes a pipeline and runs a conversation with memory:

import asyncio
import uuid

conversation_id = uuid.uuid4().hex  # Unique ID for tracking this conversation

async def main():

    conversation_pipe = await get_test_ai_handler_pipe()

    await conversation_pipe.flow(
        query_instruction="What is Agentic AI?",
        conversation_id=conversation_id  # Context maintained here
    )

    await conversation_pipe.flow(
        query_instruction="Give me real-world examples of Agentic AI applications?",
        conversation_id=conversation_id
    )

    await conversation_pipe.flow(
        query_instruction="How does memory improve AI conversations?",
        conversation_id=conversation_id
    )

if name == “main”: asyncio.run(main_conversations())

---

✅ **Supports contextual memory** per `conversation_id`
✅ **Ideal for multi-turn chat experiences** across CLI, WebSocket, or RESTful APIs

---

## 💪 Sample Output

```shell
Generate 'the response based on the user input. When you response the answer,
remember the previous context if previous context is not empty otherwise response it'.

User Input: What is Agentic AI?
Previous Context: []

    name='AI agent',
    agent_id='d03230eff9084e0097141d24289bbd2c',
reason='The provided Output_Context explains what Agentic AI is, describing its agency,
autonomy, decision-making capabilities, and ability to learn and adapt. This aligns with
the goal of getting the best results by supplying a comprehensive understanding of Agentic AI.',
result='Agentic AI refers to artificial intelligence systems that possess a form of agency,
which means they are designed to act autonomously on behalf of users or stakeholders in an
environment. This kind of AI is capable of decision-making, planning, and executing tasks
without direct human intervention, often using advanced algorithms to optimize for specific
outcomes. Unlike purely reactive or rule-based systems, agentic AI can adapt to new situations,
learn from interactions, and make complex decisions by considering various factors and potential
consequences.',
content=None,
error=None,
is_goal_satisfied=True

User Input: Give me real-world examples of Agentic AI applications?
Previous Context: []

        name='AI agent',
        agent_id='177fdd1a9704457b981bed5db33858fc',
reason='The provided Output_Context contains a clear explanation of Agentic AI, listing specific real-world applications that demonstrate AI systems
possessing agency. These examples comprehensively illustrate the autonomy and decision-making capabilities inherent in Agentic AI.',
result=[
     1.Autonomous Vehicles: Such as self-driving cars that navigate and make driving decisions without human intervention.
     2.Robotic Process Automation (RPA): Advanced systems that make autonomous decisions within business processes.
     3.Drones and Unmanned Aerial Vehicles (UAVs): Used for delivery, surveillance, or search and rescue operations.
     4.Smart Personal Assistants: Like Google Assistant, Siri, or Alexa, which make proactive suggestions or actions.
     5.Intelligent Customer Support Bots: Bots that manage complex inquiries and escalate issues autonomously.
     6.Industrial Robots: Robots in manufacturing that adapt and optimize tasks autonomously.
     7.AI in Financial Trading Systems: Systems making autonomous stock trading decisions.
],
content=None,
error=None,
verify_goal=True,
is_goal_satisfied=True

User Input: How does memory improve AI conversations?
Previous Context: []

    name='AI agent',
    agent_id='177fdd1a9704457b981bed5db33858fc',
reason='The Output_Context does not address the query about how memory improves AI conversations. Instead, it discusses Agentic AI and its applications,
which are not relevant to the query. The additional provided response outside the context clearly addresses how memory can enhance AI conversations by explaining
various aspects like context retention, personalization, learning from interactions, reducing repetition, and improved decision-making.
result="Memory can enhance AI conversations in several ways:\n\n1. **Context Retention**: Memory allows an AI to retain the context of an ongoing
conversation, enabling more coherent and contextually relevant responses. By remembering earlier interactions, an AI can more effectively follow a thread of
conversation and reference past exchanges to enhance the dialogue.\n\n2. **Personalization**: With memory, an AI can remember users' preferences, personal
details, previous questions, and feedback. This leads to personalized interactions, as the AI can tailor its responses based on individual user profiles or
specific user history.\n\n3. **Learning from Interactions**: Memory enables an AI to learn from past interactions and improve over time. As it stores information
about what worked well in previous conversations or where misunderstandings occurred, it can adapt its approach to become more effective in future
interactions.\n\n4. **Reducing Repetition**: By recalling earlier interactions, an AI can avoid repetitive questioning or information exchange, which improves
user experience. Users do not have to repeat themselves if the AI remembers what has already been discussed.\n\n5. **Improved Decision Making**: Memory allows an
AI to remember previous successes or failures, which can inform its decision-making processes. By leveraging past experiences, the AI can choose strategies that
are more likely to be effective in achieving desired outcomes.\n\nOverall, memory enhances the quality and effectiveness of AI conversations by adding continuity,
depth, and adaptability to interactions.",
content=None,
error=None,
verify_goal=True,
is_goal_satisfied=True

​
🧠 Why Memory Matters

Memory enables multi-turn conversations where the AI doesn’t forget previous questions, responses, or the conversation’s purpose.

This is essential for building helpful and coherent AI assistants that feel more natural and less robotic.