Symfony AI - Chat Component

Edit this page

The Chat component provides an API to interact with agents, it allows to store messages and retrieve them later for future chat and context-retrieving purposes.

Installation

1
$ composer require symfony/ai-chat

Basic Usage

To initiate a chat, you need to instantiate the Symfony\AI\Chat\Chat along with a Symfony\AI\Agent\AgentInterface and a Symfony\AI\Chat\MessageStoreInterface & Symfony\AI\Chat\ManagedStoreInterface:

1
2
3
4
5
6
7
8
9
10
11
12
use Symfony\AI\Agent\Agent;
use Symfony\AI\Chat\Chat;
use Symfony\AI\Chat\InMemory\Store as InMemoryStore;
use Symfony\AI\Platform\Bridge\OpenAi\Factory;
use Symfony\AI\Platform\Message\Message;

$platform = Factory::createPlatform($apiKey);

$agent = new Agent($platform, 'gpt-4o-mini');
$chat = new Chat($agent, new InMemoryStore());

$chat->submit(Message::ofUser('Hello'));

Streaming

The Chat component supports streaming responses from the LLM in real-time using the stream() method. This returns a Generator that yields DeltaInterface deltas as they are produced by the model:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
use Symfony\AI\Agent\Agent;
use Symfony\AI\Chat\Chat;
use Symfony\AI\Chat\InMemory\Store as InMemoryStore;
use Symfony\AI\Platform\Bridge\OpenAi\Factory;
use Symfony\AI\Platform\Message\Message;
use Symfony\AI\Platform\Message\MessageBag;
use Symfony\AI\Platform\Result\Stream\Delta\TextDelta;

$platform = Factory::createPlatform($apiKey);

$agent = new Agent($platform, 'gpt-4o-mini');
$chat = new Chat($agent, new InMemoryStore());

$chat->initiate(new MessageBag(
    Message::forSystem('You are a helpful assistant.'),
));

foreach ($chat->stream(Message::ofUser('Tell me a story about the sun')) as $delta) {
    if ($delta instanceof TextDelta) {
        echo $delta;
    }
}

Once the stream is fully consumed, the assistant message is automatically persisted to the message store along with the user message. This means the conversation history is kept up-to-date without any additional code.

Note

Due to implementations limitations, using streaming with MessageStore is not recommended.

Code Examples

You can find more advanced usage in combination with an Agent using the store for long-term context:

Supported Message stores

Implementing a Bridge

The main extension points of the Chat component is the MessageStoreInterface, that defines the methods for adding messages to the message store, and returning the messages from a store.

This leads to a store implementing two methods:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
use Symfony\AI\Chat\MessageStoreInterface;
use Symfony\AI\Platform\Message\MessageBag;

class MyCustomStore implements MessageStoreInterface
{
    public function save(MessageBag $messages): void
    {
        // Implementation to add a message bag to the store
    }

    public function load(): MessageBag
    {
        // Implementation to return a message bag from the store
    }
}

Managing a store

Some store might requires to create table, indexes and so on before storing messages, the ManagedStoreInterface defines the methods to setup and drop the store.

This leads to a store implementing two methods:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
use Symfony\AI\Chat\ManagedStoreInterface;
use Symfony\AI\Chat\MessageStoreInterface;

class MyCustomStore implements ManagedStoreInterface, MessageStoreInterface
{
    # ...

    public function setup(array $options = []): void
    {
        // Implementation to create the store
    }

    public function drop(): void
    {
        // Implementation to drop the store (and related messages)
    }
}

Commands

While using the Chat component in your Symfony application along with the AiBundle, you can use the bin/console ai:message-store:setup command to initialize the message store and bin/console ai:message-store:drop to clean up the message store:

1
2
3
4
5
6
7
8
# config/packages/ai.yaml
ai:
    # ...

    message_store:
        cache:
            symfonycon:
                service: 'cache.app'
1
2
$ php bin/console ai:message-store:setup symfonycon
$ php bin/console ai:message-store:drop symfonycon
This work, including the code samples, is licensed under a Creative Commons BY-SA 3.0 license.
TOC
    Version