> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/17.latest/frontend/chat-controller.md).

# Chat Controller

`UaiChatController` provides a high-level API for performing chat completions from custom backoffice elements.

## Import

{% code title="Import" %}

```typescript
import { UaiChatController } from "@umbraco-ai/core";
```

{% endcode %}

## Constructor

{% code title="Constructor" %}

```typescript
new UaiChatController(host: UmbControllerHost)
```

{% endcode %}

| Parameter | Type                | Description                                           |
| --------- | ------------------- | ----------------------------------------------------- |
| `host`    | `UmbControllerHost` | The controller host (usually `this` in a Lit element) |

## Methods

### complete

Performs a chat completion and returns the full response.

{% code title="Signature" %}

```typescript
async complete(
    messages: UaiChatMessage[],
    options?: UaiChatOptions
): Promise<{ data?: UaiChatResult; error?: unknown }>
```

{% endcode %}

| Parameter  | Type               | Description               |
| ---------- | ------------------ | ------------------------- |
| `messages` | `UaiChatMessage[]` | The conversation messages |
| `options`  | `UaiChatOptions`   | Optional configuration    |

**Returns**: Promise resolving to `{ data?, error? }`

{% code title="Example" %}

```typescript
import { LitElement } from "lit";
import { UaiChatController, UaiChatMessage } from "@umbraco-ai/core";

class MyElement extends LitElement {
    #chat = new UaiChatController(this);

    async askQuestion(question: string) {
        const messages: UaiChatMessage[] = [
            { role: "system", content: "You are a helpful assistant." },
            { role: "user", content: question },
        ];

        const { data, error } = await this.#chat.complete(messages);

        if (error) {
            console.error("Failed:", error);
            return;
        }

        console.log("Response:", data.message.content);
        console.log("Tokens used:", data.usage?.totalTokens);
    }
}
```

{% endcode %}

## Options

{% code title="UaiChatOptions" %}

```typescript
interface UaiChatOptions {
    /** Profile ID (GUID) or alias. If omitted, uses the default chat profile. */
    profileIdOrAlias?: string;
    /** AbortSignal for cancellation. */
    signal?: AbortSignal;
}
```

{% endcode %}

## Complete Example

{% code title="ai-assistant.element.ts" %}

```typescript
import { LitElement, html, css } from "lit";
import { customElement, state, property } from "lit/decorators.js";
import { UaiChatController, UaiChatMessage, UaiChatResult } from "@umbraco-ai/core";

@customElement("ai-assistant")
export class AIAssistantElement extends LitElement {
    static styles = css`
        :host {
            display: block;
            padding: 1rem;
        }
        .loading {
            opacity: 0.5;
        }
        .error {
            color: red;
        }
    `;

    #chat = new UaiChatController(this);
    #abortController?: AbortController;

    @property()
    profileAlias = "content-assistant";

    @state()
    private _messages: UaiChatMessage[] = [];

    @state()
    private _loading = false;

    @state()
    private _error?: string;

    async #sendMessage(userMessage: string) {
        // Add user message
        this._messages = [...this._messages, { role: "user", content: userMessage }];

        this._loading = true;
        this._error = undefined;

        // Create abort controller for cancellation
        this.#abortController = new AbortController();

        const { data, error } = await this.#chat.complete(this._messages, {
            profileIdOrAlias: this.profileAlias,
            signal: this.#abortController.signal,
        });

        this._loading = false;

        if (error) {
            this._error = "Failed to get response";
            return;
        }

        // Add assistant response
        this._messages = [...this._messages, data!.message];
    }

    #cancel() {
        this.#abortController?.abort();
    }

    #handleSubmit(e: Event) {
        e.preventDefault();
        const form = e.target as HTMLFormElement;
        const input = form.querySelector("input") as HTMLInputElement;
        if (input.value.trim()) {
            this.#sendMessage(input.value);
            input.value = "";
        }
    }

    render() {
        return html`
            <div class="messages">
                ${this._messages.map(
                    (m) => html` <div class="message ${m.role}"><strong>${m.role}:</strong> ${m.content}</div> `,
                )}
            </div>

            ${this._error ? html`<p class="error">${this._error}</p>` : ""}

            <form @submit=${this.#handleSubmit} class=${this._loading ? "loading" : ""}>
                <input type="text" placeholder="Ask something..." ?disabled=${this._loading} />
                <button type="submit" ?disabled=${this._loading}>Send</button>
                ${this._loading ? html` <button type="button" @click=${this.#cancel}>Cancel</button> ` : ""}
            </form>
        `;
    }
}

declare global {
    interface HTMLElementTagNameMap {
        "ai-assistant": AIAssistantElement;
    }
}
```

{% endcode %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.umbraco.com/ai-in-umbraco/17.latest/frontend/chat-controller.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.