Skip to main content
Build conversational agents that remember context across messages. This pattern maintains conversation history in memory, enabling chatbots, assistants, and any multi-turn interaction.

​
How It Works

The Dedalus SDK’s runner.run() accepts a messages array. By appending user messages and updating with result.to_input_list() after each turn, you get persistent conversations:
  1. Append the new user message to history
  2. Run the model with the full history
  3. Update history using result.to_input_list()

​
Multi-turn Chat

import asyncio
from dedalus_labs import AsyncDedalus, DedalusRunner

async def main():
    client = AsyncDedalus()
    runner = DedalusRunner(client)
    messages: list[dict] = []

    while True:
        user_input = input("You: ").strip()
        if not user_input:
            break

        messages.append({"role": "user", "content": user_input})

        result = await runner.run(
            model="openai/gpt-4o",
            messages=messages,
        )

        messages = result.to_input_list()
        print(f"Assistant: {result.final_output}\n")

asyncio.run(main())

​
Key Concepts

​
Message Format

The Dedalus SDK uses the OpenAI message format: 
[
    {"role": "user", "content": "Hello"},
    {"role": "assistant", "content": "Hi! How can I help?"},
    {"role": "user", "content": "What did I just say?"},
]

​
Persistence with to_input_list()

After each runner.run(), call result.to_input_list() to get the complete conversation history including tool calls and assistant responses. This preserves the full context for the next turn.

​
Persisting to Disk

For conversations that survive restarts, save to JSON: 
import asyncio
import json
from pathlib import Path
from dedalus_labs import AsyncDedalus, DedalusRunner

HISTORY_FILE = Path("chat_history.json")

def load_messages() -> list[dict]:
    if HISTORY_FILE.exists():
        return json.loads(HISTORY_FILE.read_text())
    return []

def save_messages(messages: list[dict]):
    HISTORY_FILE.write_text(json.dumps(messages, indent=2))

async def main():
    client = AsyncDedalus()
    runner = DedalusRunner(client)
    messages = load_messages()

    while True:
        user_input = input("You: ").strip()
        if not user_input:
            break

        messages.append({"role": "user", "content": user_input})

        result = await runner.run(
            model="openai/gpt-4o",
            messages=messages,
        )

        messages = result.to_input_list()
        save_messages(messages)
        print(f"Assistant: {result.final_output}\n")

asyncio.run(main())

​
Storage Options

StorageUse Case
In-memorySingle session, no persistence needed
JSON fileLocal development, single user
SQLiteLocal apps, moderate scale
RedisHigh-performance, distributed
PostgreSQLProduction, with JSONB columns
Connect these docs programmatically to Claude, VSCode, and more via MCP for real-time answers.
Last modified on March 10, 2026