Get instant answers to your question base on your knowledge base with the help of AI.

Updated over a week ago

What's Answer? 🔮

Answer feels magical the first time you use it. It's a totally new and efficient way to leverage your knowledge base.

Answer allows users to ask questions on the knowledge base and get an answer using artificial intelligence. Answer is available from Elium's web interface, via Slack and soon Teams.

How to use it?

✍🏻 Note

Answer is currently in beta. Apply here to activate it on your platform.

Answer is available on the web app, Slack and soon on Teams.

Answer is available from anywhere in the web application. It's located in the top right corner in the header. Click on the crystal ball to open it.

This will open a modal. The user can then ask his question.

Answer's modal.

In the example above, we can see that Answer will search through 15 different spaces to generate an answer. Type your question and click "Answer".

✍🏻 Note

Right now the platform administrator need to send an email to to change the selected spaces. This will be available in the admin panel after.

  1. Question Asked
    A reminder of the question asked.

  2. AI Generated Answer
    The most interesting part. It's the AI-generated answer to the user's question. This answer is based on the sources found in the selected spaces.

  3. Found Sources
    Those are the Stories (and soon files) used to answer the question.

    Note: Answer selects the relevant paragraphs from Stories, not the full content.

  4. Feedback
    The user has the possibility to leave a feedback. This is especially useful if the given answer can be improved or is wrong. Clicking on "Good thanks" will register the question as good. Clicking on "Not good" triggers a modal window allowing the user to explain what went wrong.

The user can pick a reason and write a comment. Those parameters are then registered and available from the admin panel (see section below "Process Feedback")

What is indexed?

Answer can read regular Stories but also files like Word and PowerPoint and even video.



First you'll need to install the Elium app from Slack marketplace.

Once installed, you'll be able to use Answer directly from Slack.

Click on the three dots menu from any message to use it as a question for Answer. Answer will send you a private message which you can then send to the channel.

In the example below, the user used the message "Is Elium secure?" as a question for Answer, which responds with a private message.

Like the web application, the user can provide feedback. If satisfied, the message can be send to the whole channel.

Answer is also available as a Slack command. Type "/answer [your question]" in the message box.

Like the first example, Answer will display a private message that you can send to the whole channel.


You'll first need to install the Elium app from Teams.

From there, you'll be able to access Answer from either the message box or the message menu.

From the message box, click on the three dots and search for Elium.

A modal will appear with a text input to type your question.

When you submit your question, Answer will create a draft message with the answer.

You can then submit this message to the channel.

The other way to use Answer directly in Teams is to use a message as a question. Click on the message menu and click "Ask Elium".


Answer can be scoped to a subset of spaces and users. The spaces used to generate the answer are stated in Answer modal.

Of course, Answer will only use the Stories available to the user asking the question to generate an answer.


Administrators can process Answer feedback from the admin panel in the "Content Management" section.

The page contains two cards :

  1. Analytics
    You can see the kind of feedback left by users at a glance. Red means negative feedback, green a positive one and grey when the user didn't leave any feedback.

  2. Feedback
    In this card you have a table listing all the unprocessed and processed feedback.


In the first card you can see the number of questions asked and the type of feedback left by the users: positive (in green) or to improve (in red).

The aim is obviously to improve things by getting more and more positive (or neutral) feedback. To achieve this, the game is quite simple: solve the feedback to fill in the knowledge gaps as you go along. That's what will do in the next section.

Like any other platform Analytics, you can select multiple period or even a custom interval. You can also export the data to analyze it in your favorite data software.

Process Feedback

Right below the analytics, you'll find a card with two tabs: Feedback and Resolved.

The first tab "Feedback" gather all you unprocessed feedback. Click on one request to display all the details in a side panel. You can browse from feedback to feedback by doing so.

The details side panel display the following request data:

  1. The question asked by the user

  2. The answer generated by Answer

  3. The sources used by Answer to generate the answer

  4. The user feedback type (positive or to improve)

  5. The date when the question was asked

  6. And the user who asked the question

There are two possibles actions:

  1. Mark as resolved: This will directly mark the feedback as resolve and will move it to the second tab "Resolved".

  2. Provide the good answer: This action will trigger an modal to write the answer yourself. You will also have the ability to link a content. This is useful when you create new content to answer the initial feedback question.

You can link content and write an answer yourself.

Export all questions

All questions asked to Answer can be exported from the admin panel from the "Answer Feedback" page. The CSV file has the following fields:

  1. Question asked

  2. Answer provided

  3. Sources used

  4. User who asked the question

  5. Date

Current Developments

Answer is under active development. Here's what we're working on at the moment:

  1. Teams integration. Ask Answer questions directly from Teams. No need to access Elium to benefit from the knowledge base!

  2. A brand new interface with the integration of the Command Center.

How can I give feedback?

You have some feedback on Answer? Perfect, that's what we're looking for! You send an email at and it will be dispatched to the product team. Thank you!

Data processing

To make Answer work, we have developed a partnership with Microsoft. We call up an API on Azure servers in Europe.

Your data remains stored on Elium servers. It transits to Microsoft Europe's servers to generate the vectors needed to run Answer's search and responses. Unlike a solution like OpenAI, which is free of charge, your data cannot be used to enhance Microsoft's artificial intelligence models.

Costs and pricing

Of course, using these models comes at a cost to Elium.

We're still considering the pricing associated with this new functionality, particularly as we don't yet have enough data to measure the impact on the cost of our infrastructure. Model access prices change regularly, and Answer is not yet widely deployed.

We'll get back to you as soon as possible.

Want to know how does it work?

It's magic 💫

When a story is indexed (created, updated, same as now) we index it semantically. The idea is to capture the meaning of the story, not just the words (as opposed to what the "regular" search does).

When the user ask a question, we search for the five paragraphs with the closest semantic similarity. And we use these paragraphs to generate a response using AI.

Did this answer your question?