Haiilo's AI feature AVA (Artificial Virtual Assistant) is a useful tool for generating sharing messages for your externally shareable posts. It can save time for your editorial team by automating the copywriting process or empowering users to create their own unique and engaging post captions.

AVA acts as a content writer who creates attention-grabbing summaries of your posts. AVA uses the post body as a reference for link and article posts and the post title as a reference for image and video posts. The language of your sharing message is determined by your platform language. When using AVA to generate sharing messages, you can:

• Select whether the message's tonality should be formal or informal. Our system prompts AVA to write formally or informally. The AI determines tonality based on its own settings.
• Select whether the message should be short or long. Our system prompts AVA to write a maximum of 120 characters or 5-10 sentences long.
• Choose whether to include emojis. Our system prompts AVA to include or exclude emojis. By default, emojis are deselected.

The system will save your selected options for future sharing messages, but you can always redefine them.

This article should not be used as a comprehensive resource for learning about artificial intelligence in general. We provide a short introduction to AI at the bottom of this article.

Set up AVA

On the AI provider

To use AVA, you require a subscription and API key from an AI provider. For this, you have to set up a contract between your company and an AI provider.

Currently, Haiilo supports the following providers. You will need to retrieve information from your account with each provider to set up AVA on Haiilo Advocacy:

On Haiilo

You need to be a Company Admin to configure the settings for AVA on Haiilo Advocacy.

1. Go to Administration > Settings > AI Assistant
2. Check Activate AVA
3. Choose an option:
   • Activate for admin users: Only Company Admins, Group Admins, or Channel Managers can use AVA to generate sharing messages. Regular users cannot use AVA to generate their own sharing messages, but they can see and use the messages that were generated by admins.
   • Activate for all users: All users can use AVA to generate sharing messages.
4. Choose your provider from the list.
5. Paste the required information in the correct fields. You can find this information in your account with the selected AI provider. The information that is required depends on the AI provider you use.
6. Select Save

Use AVA to generate a sharing message

You can use AVA to generate sharing messages when creating and sharing content.

When creating content

You can use AVA to generate sharing messages during post creation for all social networks that allow sharing messages.

1. Start creating a post as you normally would by inputting content
2. Go to the Configure step > Select sharing networks and messages section
3. Select AI under a social network's sharing message field
4. Define the message's tonality, length, and use of emojis.
5. Select Generate to create the message
6. The sharing message will be automatically input into the field when it is generated. It may take up to 30 seconds for the message to generate.
7. Review the message.
8. Continue creating additional messages for any social network. Each sharing message field has its own AI button. If an existing sharing message exists in the field, it will be overwritten with your generated message.
9. Publish the post when you're done creating it.

When sharing content

You can only create sharing messages with AVA in the sharing modal for LinkedIn and X. For all other social networks, AI-generated sharing messages can only be created when creating the post.

1. Select a social network icon under a post to open the share pop-up
2. Under the sharing message field, select AI. If an existing sharing message exists in the field, it will be overwritten with your generated message.
3. Define the message's tonality, length, and use of emojis.
4. Select Generate to create the message
5. The sharing message will be automatically input into the field when it is generated. It may take up to 30 seconds for the message to generate.
6. Review the message before you share the post.

Limitations

AI is a cutting-edge technology that is advancing rapidly through extensive research and development. There are various factors that can result in limitations and errors, including systemic biases, insufficient training data, and low-quality data, such as outdated information. Just like a human, AI sometimes makes mistakes.

This means that when using AVA, there is a risk that your generated response might be incorrect, harmful, biased, or outdated. We always recommend proofreading your AVA-generated sharing message before sharing a post to ensure its validity, integrity, and relevance. We also recommend disclosing to your targeted audience that the sharing message has been created with the help of or entirely by AI.

Privacy information

Your data is still encrypted and private following Haiilo's standard data protection policies. The AI provider's API key that you add in the Administration is stored and hosted on Haiilo servers with state-of-the-art encryption.

Your data will not be used to train OpenAI's models. OpenAI has usage policies that apply to all users of OpenAI's models and services. By following them, you ensure that AI technology is used for good. For more information, see here.

Basics of AI

AI, or Artificial Intelligence, refers to human-created technology that attempts to mimic how a human would perform a task. To do this, it uses vast amounts of data to predict results, which allows the AI to carry out tasks and provide responses based on what you ask. Like a human, it improves by learning and doing. In a nutshell, here's how it works:

• AI analyzes large datasets, known as training data, for patterns and uses this to create a model
• The model is then tested by asking questions AI already knows the answers to and assessing the accuracy of the answers. These answers are referred to as test data. 
• As the AI receives more inputs and test data, it continues to learn and further refine its model, improving the responses it provides.