Chat Components

Customizable, drop-in components for building AI-powered chat interfaces


"""LangGraph agent for the CopilotKit Showcase (FastAPI variant).Uses copilotkit's create_agent (wrapping langgraph) with CopilotKitMiddlewareso frontend-registered tools (useHumanInTheLoop, useFrontendTool) are properlyinjected into the LLM's tool list and executed on the frontend rather thanlocally."""from tools import (    get_weather_impl,    query_data_impl,    schedule_meeting_impl,    manage_sales_todos_impl,    get_sales_todos_impl,    search_flights_impl,    build_a2ui_operations_from_tool_call,)from tools.types import SalesTodo, Flightimport jsonimport timefrom typing import Anyfrom langchain_openai import ChatOpenAIfrom langchain_core.tools import tool as lc_toolfrom langchain_core.messages import SystemMessagefrom langchain.agents import AgentState as BaseAgentState, create_agentfrom langchain.tools import ToolRuntime, toolfrom langchain.messages import ToolMessagefrom langgraph.types import Commandfrom copilotkit import CopilotKitMiddlewareclass AgentState(BaseAgentState):    todos: list[SalesTodo]@lc_tooldef get_weather(location: str):    """Get the current weather for a location."""    return get_weather_impl(location)@lc_tooldef query_data(query: str):    """Query the database. Takes natural language. Always call before showing a chart."""    return query_data_impl(query)@lc_tooldef schedule_meeting(reason: str, duration_minutes: int = 30):    """Schedule a meeting. The user will be asked to pick a time via the UI."""    return schedule_meeting_impl(reason, duration_minutes)@lc_tooldef search_flights(flights: list[Flight]) -> str:    """Search for flights and display the results as rich cards. Return exactly 2 flights.    Each flight must have: airline, airlineLogo, flightNumber, origin, destination,    date (short readable format like "Tue, Mar 18" -- use near-future dates),    departureTime, arrivalTime, duration (e.g. "4h 25m"),    status (e.g. "On Time" or "Delayed"),    statusColor (hex color for status dot),    price (e.g. "$289"), and currency (e.g. "USD").    For airlineLogo use Google favicon API:    https://www.google.com/s2/favicons?domain={airline_domain}&sz=128    """    result = search_flights_impl(flights)    return json.dumps(result)@tooldef manage_sales_todos(todos: list[SalesTodo], runtime: ToolRuntime) -> Command:    """    Manage the current sales todos. Pass the full updated list.    """    updated = manage_sales_todos_impl(todos)    return Command(        update={            "todos": updated,            "messages": [                ToolMessage(                    content="Successfully updated sales todos",                    tool_call_id=runtime.tool_call_id,                )            ],        }    )@tooldef get_sales_todos(runtime: ToolRuntime):    """    Get the current sales todos.    """    current = runtime.state.get("todos", [])    return get_sales_todos_impl(current if current else None)@lc_tooldef render_a2ui(    surfaceId: str,    catalogId: str,    components: list[dict],    data: dict | None = None,) -> str:    """Render a dynamic A2UI v0.9 surface."""    return "rendered"@tool()def generate_a2ui(runtime: ToolRuntime[Any]) -> str:    """Generate dynamic A2UI components based on the conversation.    A secondary LLM designs the UI schema and data.    """    t0 = time.time()    messages = runtime.state["messages"][:-1]    context_entries = runtime.state.get("copilotkit", {}).get("context", [])    context_text = "\n\n".join(        entry.get("value", "")        for entry in context_entries        if isinstance(entry, dict) and entry.get("value")    )    model = ChatOpenAI(model="gpt-4.1")    model_with_tool = model.bind_tools([render_a2ui], tool_choice="render_a2ui")    response = model_with_tool.invoke(        [SystemMessage(content=context_text), *messages],    )    if not response.tool_calls:        return json.dumps({"error": "LLM did not call render_a2ui"})    args = response.tool_calls[0]["args"]    result = build_a2ui_operations_from_tool_call(args)    return json.dumps(result)model = ChatOpenAI(model="gpt-4o-mini")SYSTEM_PROMPT = """You are a polished, professional demo assistant for CopilotKit.Keep responses brief and clear -- 1 to 2 sentences max.You can:- Chat naturally with the user- Change the UI background when asked (via frontend tool)- Query data and render charts (via query_data tool)- Get weather information (via get_weather tool)- Schedule meetings with the user (via schedule_meeting tool -- the user picks a time in the UI)- Manage sales pipeline todos (via manage_sales_todos / get_sales_todos tools)- Search flights and display rich A2UI cards (via search_flights tool)- Generate dynamic A2UI dashboards from conversation context (via generate_a2ui tool)- Generate step-by-step plans for user review (human-in-the-loop)"""graph = create_agent(    model=model,    tools=[        get_weather,        query_data,        schedule_meeting,        search_flights,        generate_a2ui,        manage_sales_todos,        get_sales_todos,    ],    middleware=[CopilotKitMiddleware()],    state_schema=AgentState,    system_prompt=SYSTEM_PROMPT,)

Pre-built components for agentic chat#

CopilotKit's chat components give you a fully functional, customizable AI chat interface out of the box. They handle streaming, generative UI, and deep customization so you can focus on your agent's behavior, not UI plumbing.

What it looks like in code#

The live agentic-chat cell above is built from a single, small page. Wrap your UI in <CopilotKit> once (it wires the runtime, session, and agent registry) and drop <CopilotChat> wherever the chat should go:

page.tsx
    <CopilotKit runtimeUrl="/api/copilotkit" agent="agentic_chat">      <Chat />    </CopilotKit>

Inside the chat, the useConfigureSuggestions hook lets you show contextual starter prompts. The example below uses it to seed a simple "Write a sonnet" suggestion:

suggestions.ts
export function useAgenticChatSuggestions() {  useConfigureSuggestions({    suggestions: [      { title: "Write a sonnet", message: "Write a short sonnet about AI." },      {        title: "Tell me a joke",        message: "Tell me a one-line joke.",      },      {        title: "Is 17 prime?",        message: "Walk me through whether 17 is prime.",      },    ],    available: "always",  });}