AVA acts as a global search assistant, providing answers to your questions using the data available on your platform that you have access to view. Similar to the global search, AVA can access new content instantly.

To use AVA in the search, it needs to be enabled for your platform, and you need the "Search" permission.

AVA uses semantic search, which means it can understand the meaning behind a question, even if you don't use the exact wording or phrase that exists in your platform's data. For example, if you ask about "vacation days" and your company uses "holiday days" in all sources on the platform, AVA will still be able to comprehend what you mean and provide you with an answer.

Use AVA in the search

In the search, you can use AVA as follows:

1. Start a search that begins with a question word or ends with a question mark. For example, "How many self care days do we have" or "self care days?". AVA only appears when a question is asked in either of these ways.
2. Press Enter to search

An AVA Search box with AVA's answer will appear at the top of the global search. You can select Show sources to view the sources in the platform from which AVA got her answer.

Optionally, you can provide feedback on the response by answering the "Was this answer helpful?" question. The data is used solely for Haiilo's internal analytics and helps inform our improvement of AVA Search.

It's important to note that answers generated by AI may not always be accurate. We recommend carefully examining the answer and its sources to ensure its validity.

Limitations

• Content types: Answers are based only on information found in blog articles, wiki articles, content apps, and user profiles.
  • Answers will not include content from archived blog or wiki articles. This also applies to blog or wiki articles belonging to an archived community, even if the article itself is currently active.
• Source limit: AVA search is limited to retrieving a maximum of 10 sources that are relevant to answering a question. Therefore, AVA's answer should not be considered correct if a question requires consideration of more than 10 sources to be answered correctly.
• User information: AVA considers the user's permissions, job title, department, and location from their profile to more accurately identify relevant sources for each individual.
• Sorting: Content is currently organized by semantic similarity. We implement the Approximate Nearest Neighbor (ANN) algorithm to identify the most semantically similar content relevant to the user's query.
• Languages: Currently, common question words in English, German, French, and Finnish are considered. The answer is created in the language of the question.