Skip to main content

Step 4: Intelligent assistant

Ainhoa's intelligent assistants offer a powerful and straightforward way to make the most out of our data through an intuitive and accessible chat interface. With this technology, data (often complex and vast) becomes more manageable and useful. These assistants can analyze and extract relevant information from these datasets, significantly easing access to key information. This enables users to obtain precise and quick answers to their queries, streamlining decision-making processes and enhancing productivity.

The ability to conduct guided procedures is especially helpful as it allows users to follow clear and specific instructions to complete tasks or perform complex actions in a simpler, structured manner. Furthermore, integration with external systems provides access to information and functionalities from different sources, expanding the assistant's capabilities to offer comprehensive and updated solutions.

It's time to create our specialized intelligent assistant for museums. We'll consolidate all the previously done work to build and empower our bot.

Create bot

Let's go to the bots section in the side menu and click on the button.

New botNew bot

You'll be presented with a window to fill in our bot's information. We'll enter the following details:

TitleName of the intelligent assistant.ArtiBot
DescriptionBrief description of the bot. Should be entered in each available language.Allows answering questions about museums.
StrategyType of strategy the bot will use. Can be router, sequential, or LLM.Router
TagsList of tags related to the bot.
Access ModeSelects the access mode to the bot. Public access or members only.Public access
MembersList of members who will have access to the bot and their associated privilege.

Once all the necessary information is provided, click on the Create button.

Create museum botCreate museum bot

Configure bot

Once our bot is created, a detailed page will appear showcasing all the information and various configuration parameters. This page is divided into different tabs to organize all the available information.


Let's briefly explain the information available on each tab and the parameters we can adjust for our bot:

HomeProvides general information about the bot such as its title, description, or tags. Configuration options, disambiguation messages, and settings related to result interpretation and reading comprehension are available here.
GreetingsAllows us to set up greeting messages to welcome users.
MessagesEnables configuring the confidence level required to use a skill's response, the maximum number of suggestions it can provide, and a fallback message in case there's no response available.
SkillsManages all the bot's available skills for operation. Certain parameters for each assigned skill can be configured here.
User PreferencesDefault temperature values, search skills, and languages users can select in the chat.
API DetailsSettings related to accessing your bot. You can activate it, allow access via the Microsoft Bot Framework, enable telemetry, or set a limit on the message count.
Source CodeAllows visualization of your bot's entire configuration through its associated YAML source code.
Test BotEnables testing your bot specifying specific versions, identities, languages, and audiences. Additionally, you can view development information associated with each chat message. This option is quite similar to testing a skill. You can get more information here
HistoryList of all different versions your bot has.

Alright, now that we have a better understanding of the various configuration options available for our bot, let's proceed to configure ours. For our example, we'll make the following parameter changes:

  1. In the Messages tab, we set the confidence to 0.55. This indicates to the bot the minimum percentage of confidence required to respond with information provided by the skill.

Bot messages confidenceBot messages confidence

  1. In the Skills tab, we need to assign our previously created search skill.
    1. Click on the button and select museum_search.
    2. With our skill selected, we set the audience to global and establish the confidence at 0.5.

Bot skill assignedBot skill assigned

  1. In the User Preferences tab, we'll set the default options the user will use in the chat:
    1. Choose the temperature as Balanced.
    2. Select our search skill, museum_search.
    3. Choose the allowed languages, Spanish.

If we enable the Allow user to change checkbox in User Preferences, the user will be able to edit these values in the chat.

Bot user preferencesBot user preferences

  1. In the API details tab, check the Enable bot option. This allows the bot to be available for use.

Bot enableBot enable

Finally, click on Save, and now our bot is configured.

Train bot

Training our bot is a crucial and fundamental step. It involves ensuring that it's fully configured with all the parameters, skills, and services we've provided.

Training is essential for the bot to understand how to respond to different queries or commands, apply skills at the right moment, and provide accurate and relevant responses. This stage also allows it to adapt to various situations and scenarios, offering a consistent and useful user experience.

Moreover, training isn't just about learning information; it's also about understanding how to interact with users in a natural and effective manner. This involves adjusting its behavior to provide clear and friendly responses, generating a satisfying and seamless interaction.

To start the training, click on the Train button and wait for the process to complete.

Bravo! ArtiBot 🤖 is now ready to help us with our questions about museums. Let's give it a try! 💬