Accessing Symbient AI

New User Registration

To register as a new user for Symbient AI, do the following:

1. Open your web browser and navigate to the Symbient AI Home page.
2. Click on the "Sign Up" link in the upper-right section of the menu bar.
   
1. In the "Sign up" popup window,
2. Enter your first name, last name, and email in the provided boxes.
1. Enter the password you would like to use in the "Password" text box.
2. Enter the same password again in the text box displaying "Confirm password".

Note: All fields are required.

3. Click the "Signup" button to finish your registration with Symbient AI.
   1. If successful, a popup window confirming your registration will be displayed.
   2. Click "OK" to continue.
   3. A confirmation email titled "New User Confirmation Request from Symbient" will be sent to the email address provided to confirm your email address.
   4. Open the email and click the Click to Confirm button
   5. A new page will open that will display the following message:
   1. "Your email address has now been confirmed. Please sign in to continue."
   6. Click the "Sign In" link to proceed to the User Login.

Logging In

To log in as a user of Symbient AI, do the following:

1. Open your web browser and navigate to the Symbient AI Home page.
2. Click on the "Already a member?" link found in the upper-right section of the menu bar.
3. In the "Login" popup window:
   1. Enter an email address in the text box displaying "Email".
   2. Enter a password in the text box displaying "Password".
4. Click the Login button to finish logging into Symbient AI.
   1. If successful, the Symbient AI Home page will be displayed.

Brain (Home)

The Symbient AI Home page is the Brain, this page includes a chat window that can be used to interact directly with Symbient AI. Here, you can enter a request in the text box at the bottom of the screen and when clicking "Send", the text is processed by Symbient AI using the currently selected LLM Model and AI Actor.

To get to the Brain, click on the Home icon in the top of the left-hand Menu Bar.

Brain Details

The Brain page shows the following options at the top of the page:

• Model: This is the currently used LLM Model for the chat. Only Models with the "LLM" Model type that are active will be shown here. For more information about LLM Models, please see Configuring an LLM Model.
• Show data sources?: This can be toggled on and off to instruct Symbient AI to show what specific documents were referenced when inferring its reply. For more information about Data Sources, please see the Data Sources Tab.
• Take on the role of: This is where an AI Actors role can be assigned to the Model. This role will integrate the instructions of the AI Actor into the request and final response. For more information about AI Actors, please see Configuring AI Actors.

Toolbox

The Symbient AI Toolbox is used to set up and view the many different tools available within Symbient AI.

To get to the Toolbox, click the Toolbox icon on the left-hand Menu Bar.

The following is a quick overview of the tools available in the Toolbox. For more information about each of these, see the sections below.

Activities

Activities are predefined actions or instructions that can be run within other areas of Symbient AI. An Activity gives a set of instructions to an LLM Model. These instructions can be any natural language directions such as "Summarize this transcript" or "Retrieve all emails mentioned in this text."

To get to the Activities page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Activities card inside the Toolbox.

Configuring Activities

1. Add a new Activity by clicking the Add button in the top-left of the Activities window, or select an existing Activity from the Activity table.
2. On the Activity page, fill in the following information:
   • Name - A name used to identify your Activity.
   • LLM Model - The LLM Model that your Instructions will be given to. Only "active" LLM Models will appear here.
   • Is Active? - Check this box to have the Activity appear in the list of selectable Activities.
   • Description - A description of this Activity. This is not included in the instructions passed to the LLM Model.
   • Instructions - The instructions guide the AI's actions, focusing on "what" the LLM Model should do.
3. Save the current configuration settings by clicking the Save button in the bottom-right corner of the Activity window.

Cloning Activities

Activities can be cloned to quickly create a duplicate of a set of instructions.

To clone an Activity from the Activities page:

1. Click the checkbox to the left of the Activity you wish to clone. Only one Activity can be cloned at a time.
2. Click on the Clone button in the top left of the window.
3. Enter in the name for the new Activity into the "Cloned Activity Name" field.
4. Click Submit to create the new Activity.

Deleting Activities

Activities can be deleted to remove them from the database. Any Activities deleted this way cannot be recovered and the Instructions associated with the Activity will be lost forever.

To delete Activities from the Activities page:

1. Click the checkbox to the left of the Activity you wish to delete. Multiple Activities can be selected and deleted at a time.
2. Click on the Delete button in the top left of the window.
3. Click Yes to delete the selected Activities.

Running Activities

Activities can be run from two places within Symbient AI.

• Within an Automation chain with the Activity Operator.
• Used directly on a Transcription from the Activities tab.

See the sections mentioned above for specific information on utilizing Activities.

AI Actors

AI Actors are sets of additional instructions that can be added to the base model to refine its response mechanism. These Actors are designed for seamless interchangeability, allowing for flexible adaptation during user interactions with the application. AI Actors must be configured in the Toolbox before they can be used within the app.

To get to the AI Actors page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the AI Actors card inside the Toolbox.

Configuring AI Actors

1. Add a new Actor by clicking the Add button in the top-left of the AI Actors window, or select an existing Actor from the Actors table.
2. On the New Actor page, fill in the following information:
   • Name - This is the only required field.
   • Is Active? - Check this box to have the Actor appear in the list of selectable Actors on the AI Actors homepage.
   • Description - A description of this Actor. This is not included in the instructions passed to the LLM.
   • Prompt - The "Prompt" section outlines instructions guiding the AI's actions, focusing on "what" the AI should do. These instructions will influence Symbient’s responses.
   • Personas - The "Personas" section adds personality to Symbient AI, influencing "how" the AI communicates. It can be used to provide Symbient AI Models with character traits and preferences, such as age or hobbies, to make interactions more engaging and natural.
3. Save the current configuration settings by clicking the Save button in the bottom-right corner of the Actor window.

Deleting AI Actors

1. Select the Actors to be deleted by checking the checkboxes next to the target Actors.
2. Click the Delete button in the top left of the window.
3. Click Yes to delete the selected Actors.

Automations

Automations allow you to create automated actions that can perform a wide variety of functions. These Automations use individual steps called Operators, which can execute tasks such as reading or sending emails, running an Activity, passing text to an LLM Model, or reading and writing data to a webhook. These Operators are chained together to form an Automation Chain.

To get to the Automations page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Automations card inside the Toolbox.

Configuring Automations

1. Add a new Automation by clicking the Add button in the top-left of the Automations window, or select an existing Automation from the table.
2. On the Automation Details page, fill in the following information:
   • Name - A name used to identify your Automation.
   • Automation Is Active? - Check this box to activate the Automation and allow it to be triggered. How the Automation will be triggered will be controlled by the Trigger Operator that this Automation starts with.
   • Description - A text description of this Automation to help users quickly understand what its function is.
3. Save the current configuration settings by clicking the Save button in the bottom-right corner of the Automation Details window. When creating a new Automation, the details page must be saved before you can access the Flow Tab and History Tab.

Important: Please note that our servers are using GMT time for any scheduling related actions. This helps keep everything standardized between servers, users and our team.

Flow Tab

The Flow tab is where the Automation Chain for the Automation is set up. This controls the logic for the Automation which will perform its actions in order from top to bottom. There are two kinds of Operators: Triggers and Actions. Each Automation Chain will start with one Trigger Operator and then follow a series of Action Operators.

• In the Flow tab, a list of Operators will be available on the left side of the screen. To hide or show this list, click the double arrow tab near the top of the Operators panel.
• To use an Operator, drag it from the Operators panel onto the Automation Chain view screen. Once it is placed, the properties panel for this Operator will open from the right side of the window. For more information about this panel, please see Operator Properties.
• The buttons to the right of the Operators panel can be used to control the view window for the Automation Chain. The "+" and "-" buttons will zoom in and out of the view window, and the box button will reset the view to the closest zoom possible.

Trigger Operators

Trigger Operators are used to start an Automation Chain. They can involve reading data from a database, webhook, or email, or running at a scheduled time. Only one Trigger Operator can be active on an Automation Chain at a time.

The available Trigger Operators are:

• AWS S3 Bucket - Activate when an object is added or deleted from an S3 Bucket.
• Read Email - Activate when an email arrives at the specified address.
• Schedule - Activate at a specific time, date, or weekly schedule. The time provided will be run based on the GMT timezone.
• Transcription - Access a saved Transcription.
• Webhook - Activate when a webhook endpoint is called.

Action Operators

Action Operators are used to continue the Automation Chain after the Trigger Operator is activated. Actions include a wide variety of functions such as running an Activity, scraping a website, transcribing audio, or sending an email.

The available Action Operators are:

• Activity - Perform a predefined Activity from the list of currently active Activities.
• Add/Map Fields - Create one or more Output Fields populated with data that can be used by other Operators.
• Aggregator - Combine data from multiple Operators into a single JSON object that can be passed to other Operators.
• AWS S3 DELETE - Delete an object from an AWS S3 bucket.
• AWS S3 GET - Get an object from an AWS S3 bucket.
• AWS S3 PUT - Update an object within an AWS S3 bucket.
• Call Webhook - Send a JSON object to a webhook endpoint.
• Classify - Run a Classification Model on the given text.
• Fuzzy Matcher - Perform a fuzzy match comparison between two inputs.
• HTTP GET - Submit a GET request to the specified address.
• HTTP POST - Submit a POST request with a payload to the specified address.
• HTTP PUT - Submit a PUT request with a payload to the specified address.
• JSON Parser - Parse JSON data to be used for other Operators.
• LLM - Send text to an LLM Model for inference.
• Save Draft - Save an email draft.
• Scrape Website - Scrape a website for text data.
• Send Email - Send an email using an email SMTP server.
• Transcribe Audio - Send an audio file to be transcribed by Symbient ASR.
• Transcription History - Append an output to the Activity tab of a Transcription.
• Transcription Lookup - Access a Transcription that has already been processed.
• Transcription Update - Save a Transcription and update the text.

Operator Properties

When you select an Operator that is within your Automation Chain, a Properties panel will open on the right side of the window. This panel is used to control the settings of the selected Operator. Here is a quick overview of the tabs within the Operator Properties panel:

• Settings Tab - The Settings Tab is used to enter fixed values for the properties for an Operator.
• Field Map Tab - Here you can set a property from the Settings tab to use any generated output from an Operator higher up on the Automation Chain. When the property is set here, it becomes locked on the Settings Tab.
• Output Fields Tab - This is where you can see the available output fields that this Operator will provide along with the names of the fields.

History Tab

The History tab can be used to view any time the current Automation was triggered or tested.

• To see more details about an entry, click on the green "+" button on the left side of the table. This will show a breakdown of each Operator that ran and the status of the Automation at that time.
• Press the Refresh button in the top left of the window to retrieve any recent instances of the Automation being triggered or tested.

Getting Started with Your First Automation

In this guide, we’ll walk you through setting up an example automation. This example involves creating a system that will email you a message saying good morning at the start of your day! Follow the steps below to build your first Automation Chain:

1. Navigate to the Automations page.
2. Create a new Automation. For more information about the initial Automation setup, please see Configuring Automations.
3. Click on the Flow tab at the top of the window.
4. To start with, we will drag the Schedule Trigger from the Trigger Operators list over into the Automation Chain window.
5. The properties panel for the Schedule Operator will open on the right side of the window. Go to the Settings tab within this window and fill in the following fields.
   • Time - Set the time to "7:00 am" or whatever value you prefer.
   • Days - Select the "Weekdays" option to have it run Monday-Friday.
6. Click the Apply button within the Properties window to save the Operator.
7. Next, we go back over to the Operators panel and click on the Actions tab at the top of the window.
8. Find the LLM Operator within the Actions list. Drag the LLM Operator onto the Schedule Operator to connect it, a blue circle should appear to indicate that a connection will be formed.
9. The properties panel for the LLM Operator will open on the right side of the window. Go to the Settings tab within this window and fill in the following fields.
   • Prompt - Set the prompt for the LLM here, for example, "Write a unique good morning message."
   • Model - Select the "Base Model" or another LLM of your choice.
10. Click the Apply button within the Properties window to save the Operator.
11. In the Operators panel to the left, find the Send Email Operator within the Actions list. Drag the Send Email Operator onto the LLM Operator to connect it, a blue circle should appear to indicate that a connection will be formed.
12. The properties panel for the Send Email Operator will open on the right side of the window. Go to the Settings tab within this window and fill in the following fields. These provided values are just examples, you can use your own SMTP login instead.
    • Server Name/Address - The server address for your email. For Gmail this would be "imap.gmail.com".
    • Server Port - The server port. For Gmail IMAP, this would be "993".
    • Server Type - The server type. For Gmail, this would be "IMAP".
    • Connection Security - The server’s security type. For Gmail, this would be "SSL/TLS"
    • Username - Your username to login to the server.
    • Password - Your password to login to the server.
    • From - The email address that this email will be sent from.
    • To - The email address that this email will be sent to.
    • Subject - Subject of the sent email.
13. For the Body of the email, we will go to the Field Map tab of the Operator Details window. Scroll to the bottom of the window and click on the dropdown menu for Body. Click on "Activity (Result)" to link the activity output to this field.
14. Click the Apply within the Properties window button to save the Operator.
15. Click the Save button to save the Automation Chain.

Now you should receive a good morning message each day from Samaritan! To test this message, click the Test button in the bottom right corner of the window.

Operator Documentation

Operators are the building blocks of an Automation. They are the individual steps within an Automation Flow, each designed to perform a specific function or task. Operators can either trigger actions, process data, or interact with external systems, making them essential for creating effective and dynamic Automations. For more information about the usage details of each Operator, please see our Operators Documentation.

LuLa Bot

The LuLa Bot page shows the settings and output for a customizable LulaBot plugin. This plugin can be customized and added to any webpage to enable LuLa conversations on any platform.

To get to the LuLa Bot page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the LuLa Bot card inside the Toolbox.

LuLa Bot Details

To customize the LuLa bot, fill in these fields:

• Bot Name - This will change the name shown at the top of the chat window.
• Icon URL - This controls the icon shown on the widget button that will open the chat window.
• Model - This dropdown shows a list of all LLM Models that are currently active within the user’s group. This controls the "modelid" that the Source Code references.
• Greeting Placeholder - This controls the placeholder text shown in the chat window where users enter their message. This sets the tone for the intent of this chatbot.
• Color Choices - These color pickers change the background and font colors of the chat window.
• Your API and ENC keys will automatically be populated into the "chatbot" object in the Source Code window.

To utilize this chat window on a webpage, copy the HTML code from the Source Code section into your webpage’s HTML. For more information, please see the Integrations Tab.

Models

A model in the context of Large Language Models (LLMs) is a system trained to process, understand, and generate human language. They function as a way to switch between different use cases and specializations within Symbient AI applications. There are three primary classifications of Symbient AI models, each designed for distinct purposes:

1. LLM (Large Language Model): An LLM Model can pull information from large Data Sources and utilize a specialized Vocabulary. The primary function of an LLM is to interpret user requests and generate responses by leveraging its vast repository of text information.
2. ASR (Automatic Speech Recognition): An ASR Model is used when processing Transcriptions. The main function of the ASR Model is to help Symbient recognize specific words or phrases it might not know, such as technical terms or industry-specific language, like medication names or legal jargon.
3. Classification: The Classification model is used to analyze user requests and identify Intents. It is particularly useful when trying to classify different types of user engagements, such as distinguishing positive from negative feedback or identifying requests for transactions, such as placing or modifying orders.

To get to the Models page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Models card inside the Toolbox.

Adding a New Model

To add a Model:

1. Click on the Add button on the top-left corner of the window. This will take you to the Model details page.
2. Fill in the following required details:
   • Name
   • Model Type - For more information on which type of Model to select, see Models
     • Warning: The Model Type cannot be changed after the Model is created.
3. These details are not required:
   • Description - Description of this Model
   • Tags - Tags help to differentiate between models, making them easier to filter, find, and select from the list on the main Models page.
4. Click the Save button in the bottom right corner of the window to save the Model to the database.
   IMPORTANT: Save the model before continuing to the other tabs! The details in the other tabs need the model to be saved before they can be filled in.

Depending on the Model Type that is selected, different detail tabs will be shown. See the sections below for how to configure each type of Model.

Configuring an LLM Model

To configure an LLM Model:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Models card inside the Toolbox.
3. Select the target LLM Model from the list of Models.
4. For information about the Details tab, see Adding a New Model.
5. Continue with the Data Sources and Vocabulary sections below.

Data Sources Tab

Data Sources act as a teaching tool for the LLM, providing it with new information. This can include any text or PDF documents, and website URLs that the LLM can search through to find relevant information for the current interaction. Examples of Data Sources include help documentation describing how to use an application, or a webpage with useful information about a company, its products, office hours, locations, and so on. By adding Data Sources, the LLM can give more informed and accurate responses, drawing from the provided knowledge.

Adding an LLM Data Source

There are two ways to add Data Sources to your LLM:

1. To add a file, drag and drop the file into the "Drop Files Here" area or click on this area to open a file selection window.
2. To add a web page, enter the full web page URL into the URL field and press the Import button.

Deleting an LLM Data Source

To remove a Data Source from an LLM:

1. Find the target Data Source in the Data Sources tab.
2. Click the red "X" button next to the Data Source name to remove it from the Model.

Vocabulary Tab

The Vocabulary section helps the LLM recognize specific words or phrases it might not know, such as technical terms or industry-specific language, like medication names or legal jargon. Introducing these terms to the LLM helps it accurately understand and respond to more specialized queries. For instance, provided vocabulary can help Symbient AI correctly identify different similar-sounding terms, enhancing its ability to provide precise transcriptions of conversations.

Adding LLM Model Vocabulary

There are two ways to add terms to the Model’s Vocabulary:

1. To add a file, drag and drop the file into the "Drop Files Here" area or click on this area to open a file selection window.
2. To directly add terms to the Vocabulary,
   1. Add the term within the Text field on the upper right side of the window.
   2. Add a definition or description of this term within the Description field.
   3. Click the Add button below these fields to save the term.

Removing LLM Model Vocabulary

To delete a term from the LLM Model’s Vocabulary:

1. Find the target Vocabulary term in the Vocabulary tab.
2. Click the red "X" button next to the Vocabulary "Value" to remove it from the Model.

Configuring an ASR Model

To configure an ASR Model:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Models card inside the Toolbox.
3. Select the target ASR Model from the list of Models.
4. For information about the Details tab, see Adding a New Model.
5. Continue with the Vocabulary section below.

Vocabulary Tab

The Vocabulary section helps the ASR recognize specific words or phrases it might not know, such as technical terms or industry-specific language, like medication names or legal jargon. Introducing these terms to the ASR helps it accurately understand and respond to more specialized queries. For instance, provided vocabulary can help Symbient AI correctly identify different similar-sounding terms, enhancing its ability to provide precise transcriptions of conversations.

Adding ASR Model Vocabulary

There are two ways to add terms to the Model’s Vocabulary:

1. To add a file, drag and drop the file into the "Drop Files Here" area or click on this area to open a file selection window.
2. To directly add terms to the Vocabulary,
   1. Add the term within the Text field on the upper right side of the window.
   2. Add a definition or description of this term within the Description field.
   3. Click the Add button below these fields to save the term.

Removing ASR Model Vocabulary

To delete a term from the ASR Model’s Vocabulary:

1. Find the target Vocabulary term in the Vocabulary tab.
2. Click the red "X" button next to the Vocabulary "Value" to remove it from the Model.

Configuring a Classification Model

To configure a Classification Model:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Models card inside the Toolbox.
3. Select the target Classification Model from the list of Models.
4. For information about the Details tab or about adding a new Classification Model, see Adding a New Model.
5. Continue with the tab sections below.

Intents Tab

Intents are placeholders that give a name to the output data from the Slots and Input Tabs. The names given to these Intents help Symbient AI to understand what the user wants to accomplish or what information they are seeking. The "name" of the intent can also match a function name found in Symbient AI’s custom plugins. An example of this might be "weather.get" being the intent name which matches the "weather.get" function in a custom Symbient AI plugin.

Adding an Intent

To add an Intent:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Intents tab at the top of the window.
3. Click on the Add button on the upper-left side of the window.
4. Enter in the Name of the new Intent. This name can be changed at any time.
5. Click Save to add the Intent to the Intents list, or click Dismiss to return to the Intents list.

Deleting an Intent

When an Intent is deleted, all corresponding connections to any Inputs or Slots will be broken. These Inputs and Slots will need to be reassigned to a different Intent to be active again. To delete an Intent:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Intents tab at the top of the window.
3. Check the checkboxes of any Intents that should be deleted.
   • Multiple Intents can be deleted at once.
4. Click on the Delete button on the upper left side of the window.
5. Click Yes to confirm deleting these Intents, or click No to cancel the deletion.

Slots Tab

Slots are predefined variables within a conversational model that are used to extract and process specific pieces of information from user input. These variables allow the model to understand and act upon the details of a user's request. For instance, in the context of a customer seeking to book a flight with an airline, the model could have Slots for travel parameters such as the departure city, destination, and travel date.

Adding And Configuring Slots

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Intents tab at the top of the window.
3. Click on the Add button on the upper-left side of the window, or click on an existing Slot’s name to edit that Slot.
4. The following fields are required:
   • Name - The display name for this Slot.
   • Intent - This will be the Intent that this Slot applies its logic to. The Intent value will be controlled by this Slot.
5. Check the "Is Required?" checkbox to ensure that the slot will be filled before the Model continues.
6. Add any Conditions that will limit this Slot.
   • An example of a Condition would be getting someone’s name when they introduce themselves. To do this we can set the Location to "After" and Value 1 to "My name is". This will grab the text after "My name is" and apply it to this Slot.
7. Add a "Follow-up Question" if necessary.
   • When the Model is attempting to fill in a required Slot, it will use this Follow-up Question to ask the user to provide the necessary data.
8. The Description is used for informational purposes only and does not affect the Model.
9. Click Save to add the Intent to the Intents list, or click Dismiss to return to the Intents list.

Deleting Slots

Deleting Slots will break any connections that have been established with Intents through the use of the double curly-bracket naming system {{slotName}}. These connections can be reestablished by simply renaming a new Slot to the same Name. To delete a Slot:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Intents tab at the top of the window.
3. Check the checkboxes of any Slots that should be deleted.
   • Multiple Slots can be deleted at once.
4. Click on the Delete button on the upper left side of the window.
5. Click Yes to confirm deleting these Slots, or click No to cancel the deletion.

Import Tab

The Import tab can be used to import large quantities of Inputs at once. The text entered into the Import box will be broken up depending on the Type selected.

Importing Inputs

To import Inputs:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Import tab at the top of the window.
3. Select the Type of Import. For more information about Types, see Import Types.
4. Enter the Input text into the textbox below Types.
5. Click on the Import button on the upper left side of the window.

Import Types

• Plain Text - The Plain Text selection will break up the text into individual sentences and add each sentence as a new Input row.
• Intents (CSV) - Each row in the file is added as a new row on the Input tab where the first column in the row is the "Input", the second column is the "Language", and the third column is the "Intent". The Intent specified will have to match one of the intents listed on the Intents Tab. Because of this, it is best to have your Intents created before importing a CSV file.

Input Tab

The Inputs tab allows users to assign expected user inputs with an Intent. These example Inputs are used to train the Model and can be used to pull data from user inputs to enable the AI to perform Actions.

Adding an Input Row

To add an input row:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Inputs tab at the top of the window.
3. Click on the Add Row button on the upper-left side of the window.
4. The new row will immediately be added to the top of the Inputs list.

Configuring an Input

To configure an Input after adding an Input row:

1. Enter in the text that the model should be trained on in the left-hand textbox.
2. Select the Language for this Input, the default is en-US for US-English.
3. Select the Intent that this Input should be identified as.
   • Leaving an Intent blank allows the Model to use "No Classification" as one of its options.
   • If every Input has an Intent assigned, the Model will attempt to apply an Intent classification to every section of the text.

• An Input can be removed by pressing the "X" button to the right of the Intent selector.
• All inputs can quickly be assigned values for the Language and Intent fields by using the drop-down selectors at the top of the Inputs table. WARNING: This will overwrite ALL Inputs.
• If you overwrite some data and want to return to the previously saved values, refresh the page to undo any changes since the last save.

Test Tab

The Test tab for Classifications allows users to test the current configurations of their Model. Any tests sent here will be processed by the current Trained version of the Model. If the Model has not been recently Trained, the responses may not be using the most recent configuration settings.

Testing A Classification

To test a Classification Model on the Test tab:

1. Go to the Classification Model page of the Model to be edited.
2. Click on the Test tab at the top of the window.
3. Enter the test message within the textbox on the top half of the window, above the Test and Reset buttons.
4. Click the Test button to run the Classification Model on the provided text.
5. The output will appear in the text box below, any text that was identified by the Classification will be highlighted in the output.

Classification Model Details

Classification Models must be deployed before they can be used within the app or in a public setting. To get a Classification Model deployed, the model must first be trained. This training data can be exported and imported to make adjustments quickly or to save past configurations.

Training a Classification Model

To train the Classification Model:

1. Go to the Classification Model page of the target Model.
2. Configure all of the Model’s tabs.
3. Once the Model has been configured, click the Train button on the bottom right side of the window.
4. A popup should alert you when the training has been completed.
   • If the training fails despite the Model’s configuration being correct, please contact Symbient AI support.
5. Once the Model is trained, its function can be tested on the Test Tab.

Exporting a Classification Model

Model data can be exported and saved locally by the user as a backup or to quickly switch between different configurations for the same Model. The exported Model data can be imported on the Import Tab.

To export a Classification Model:

1. Go to the Classification Model page of the target Model.
2. Click on the Export button in the bottom right corner of the window.
3. This should start a download of the Model’s configuration and training data within a .zip file.
   • The data.json file contains the base configuration information.
   • The model.json file contains the trained model data.

Deploying And Undeploying a Classification Model

Deploying a Classification makes that Model available as an option to any Classification Model selection dropdowns.

To Deploy a Classification Model:

1. Go to the Classification Model page of the target Model.
2. Click on the Deploy button in the bottom right corner of the window.

To Undeploy a Classification Model:

1. Go to the Classification Model page of the target Model.
2. Click on the Undeploy button in the bottom right corner of the window.

Transcriptions

Transcriptions are text representations of audio files. Transcripts within Symbient are either uploaded as audio and then transcribed or the text file for the transcript can be uploaded directly. The directions for these two processes are detailed below.

To get to the Transcriptions page:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Transcriptions card inside the Toolbox.

There are two ways of creating a Transcription within Symbient:

1. Uploading the Audio to be processed by Symbient ASR
2. Uploading the Transcription text directly to Symbient

Uploading Audio To Be Processed Into a Transcription

To upload audio on the Transcripts page:

1. Click the Upload Audio button at the top of the window, this will open a file selection window.
2. Select the audio file to be uploaded.
3. The Upload Audio window will open with the following fields:
   • Reference Label - This will be the reference name for the file.
   • Tags - These tags can be used to organize Transcripts.
   • Description - This can be used to describe the Transcript.
4. A progress bar will display the upload progress.
5. Once the audio has been uploaded, it will appear in the Transcripts list. The Status will be set to Processing initially.
   • Once the audio has been processed the status will change to Complete.
   • If there are any issues with the processing, the status will show as NA.

Uploading Transcription Text Directly

To directly upload Transcription text:

1. Click the Import Transcription button at the top of the window, this will open a file selection window.
2. Select the text file to be uploaded.
3. The Import Transcription window will open with the following fields:
   • Reference Label - This will be the reference name for the file.
   • Tags - These tags can be used to organize Transcripts.
   • Description - This can be used to describe the Transcript.

Downloading Multiple Transcription Texts or PDFs

To download Transcriptions as text or PDF files:

1. Check the checkboxes of Transcriptions you want to download from the Transcriptions list.
   • Multiple Transcriptions can be selected to combine them into a single file.
2. Click the Download / Export Selected button.
3. This will open the selected Transcriptions in another tab on your browser in the selected file format.
   • To quickly save a text output, click within the window of the text and press CTL+A to select all of the text and CTL+C to copy the text.
   • To save a PDF output, press the Download button in the top right corner of the window.

Deleting Transcriptions

To delete a Transcript from the database:

1. Check the checkboxes of Transcriptions you want to remove from the Transcriptions list.
2. Click the Delete Transcriptions button to remove the selected Transcriptions.
   • WARNING: Exercise caution when doing this. Transcriptions deleted from the database cannot be recovered.

Transcription Details Tab

To reach the Transcription Details page for a specific Transcript:

1. Click on the Toolbox icon on the left-hand Menu Bar to get to the Toolbox.
2. Click on the Transcriptions card inside the Toolbox.
3. In the Transcriptions list, click on the Transcription ID or Reference Label of the target Transcription.
   • This Transcriptions list can be filtered by any column within the Advanced Filters section.
4. The following fields can be adjusted for the selected Transcript:
   • Reference Label - The reference label for a Transcript cannot be empty.
   • Status - The current status of this audio upload, can either be "Processing", "Complete", "Failed" or "NA".
   • Audio File URL - The URL of the audio file.
   • Tags - Tags can be added to help organize transcriptions.
   • Description - A description of the audio upload.
   • Transcript - This section of the window will show the text output of the audio once it is processed.
   • Speakers - This can be used to set the names of any speakers identified within the Transcription.

Changing Speaker Names Within a Transcript

To change the names of Speakers within a Transcript:

1. Go to the Transcription Details page.
2. Select the target speaker from the dropdown menu next to Speakers.
3. In the text field to the right of the dropdown, enter the name that should be applied to this speaker.
4. Press the Checkmark button to the right to apply the label.

Downloading Transcription Text or PDF

To download a Transcript as a text or PDF document:

1. Go to the Transcription Details page.
2. Click on the Download / Export button in the bottom left corner of the window.
3. Then select either As Text or As PDF.
4. This will open the selected Transcript in another tab on your browser in the selected file format.
   • To quickly save a text output, click within the window of the text and press CTL+A to select all of the text and CTL+C to copy the text.
   • To save a PDF output, press the Download button in the top right corner of the window.

Running Classification Models On A Transcript

To run a Classification Model on a Transcript:

1. Navigate to the Transcription Details page for the target Transcript.
2. Click on the Classifiers tab at the top of the window.
3. Next to Select Classifier in the bottom right corner of the window, click the dropdown menu to select the Classification Model that should be run on this Transcript.
4. Click on the Run button to process the text.
5. Within the Classifiers window, two additional windows will appear.
   • The window on the right shows the Intents that this Classification Model is looking for within the text. A quick breakdown of the number of occurrences and percentage of text for each Intent is shown to the right.
   • The window on the left shows the Transcript text highlighted with the corresponding Intent.
6. Click the Reset button to clear these windows.

Running Activities On A Transcript

Activities can be run on a Transcript to process the Transcript text through an LLM Model with the instructions specified within that Activity. In the event of selecting several Activities to run, they will be queued up and run asynchronously in the background. So, some Activities may show as "Processing" on the Activities tab and will be updated once they've been completed in the LLM.

To run an Activity on a Transcript:

1. Navigate to the Transcription Details page for the target Transcript.
2. Click on the Activities tab at the top of the window.
3. Next, to Select Activity in the bottom right corner of the window, click the dropdown menu to select the Activity that should be run on this Transcript.
4. Click on the Run button to process the text.
5. Within the Activities window, the output from the activated Activity will appear.
   • The underlined text is the name of the Activity that was activated.
   • The text below is the output from that Activity.
6. Multiple Activities can be run on a text and can be displayed at the same time on this page.
7. Click the Reset button to clear this page and remove all outputs.

Users

The Users page allows you to manage any users within your Team. This includes adding new Team members, editing team member details, and deleting Team member accounts. You must have the Company Admin or User Manager Roles to access the Users page.

To access Users (other Team Members), click on the Team icon on the left-hand Menu Bar.

If you have an Enterprise plan and you have Single Sign-On (SSO) enabled, please see our SSO Documentation for more information about configuring and accessing Symbient AI via SSO.

To add new Team Members to Symbient AI do the following:

1. Click on the Team icon on the Menu Bar.
2. Click the Add button. This will open the User Details tab.
3. Enter the new member’s Login Info. The following fields are required:
   • Email Address
   • First Name
   • Last Name
   • Password
     • The password must follow the Symbient AI Password Policy.
     • Click on the Eye icon to the right of each password field to show or hide the password text.
4. Select optional fields.
   • User is active?
     • Checked means that the User is Active. An inactive user will not be able to log in.
   • User uses 2FA?
     • Checked means that the User will be required to verify with 2-Factor Authentication when logging in. For more information about 2FA, please see 2-Factor Authentication Tab.
   • User Roles - Check the boxes for any role that should be applied to this user. For more information about specific Roles, please see the Roles section within the Glossary.
5. Click Save when finished.
6. After being prompted that a Team Member has been added, click Dismiss to return to Users.

Maintaining Team Members

To edit existing Team Members within Symbient AI do the following:

1. Click on the Team icon on the Menu Bar.
2. Click on the Name of the user to edit.
3. This will open User Details tab.
4. The following fields are required for the edits to be saved:
   • Email Address
   • First Name
   • Last Name
5. Changing the password is not required. If you do not want to change this user’s password, leave these fields blank.
   • Password
     • The password must follow the Symbient AI Password Policy.
   • Confirm Password
6. Select optional fields
   • User is active?
     • Checked means that the User is Active. An inactive user will not be able to log in.
   • User uses 2FA?
     • Checked means that the User will be required to verify with 2-Factor Authentication when logging in. For more information about 2FA, please see the 2-Factor Authentication Tab.
   • User Roles - Check the boxes for any role that should be applied to this user. For more information about specific Roles, please see the Roles section within the Glossary.
7. Click Save when finished.
8. After being prompted that the current Team Member settings have been updated, click Dismiss to return to Users.

Deleting Team Members

To delete Team Members in Symbient AI do the following:

1. Click on the Team icon on the Menu Bar.
2. Select the Team Members to remove by checking the checkbox to the left of the Team Member’s name.
3. Click the Delete button in the top left corner of the Team Members window.
4. Click Yes when prompted to Confirm.
   • To cancel Delete, Click No.

Managing Account Settings

The Account Settings section allows users to modify their profile details. Here, users can update various settings, including their email address, password, two-factor authentication, organization, membership type, payments, payment details, and integrations.

To reach the Account Settings page, click on the Settings icon on the menu bar.

Details Tab

The Details tab is used for updating Individual User Settings. You can change your First Name, Last Name, and email address, and turn on or off 2 Factor Authentication.

Change Detail Information

To change your First Name, Last Name, or Email address, do the following:

1. Click on the Settings icon on the Menu Bar.
2. This should take you to the Details tab. If it does not, click on Details.
3. The following fields are required for the edits to be saved:
   • Email Address
   • First Name
   • Last Name
4. Click Save when finished.

Enable Or Disable Two Factor Authentication (2FA)

Two factor authentication can be used to provide more security for your account. For more information about 2FA, please see the 2-Factor Authentication Tab section below.

To enable or disable 2FA, do the following:

1. Click on the Settings icon on the Menu Bar.
2. This should take you to the Details tab. If it does not, click on Details.
3. Check or Uncheck the box labeled: Use 2-Factor Authentication for added security?
   • Please ensure that 2-Factor Authentication is already configured for this account before enabling this. For more information about 2FA, please visit 2-Factor Authentication.
4. Click Save when finished.

Change Password

To change your Password, do the following:

1. Click on the Settings icon on the Menu Bar.
2. This should take you to the Details tab. If it does not, click on Details.
3. Click Update Password on the lower left of the screen.
4. In the Password window:
   • Enter the current password in the Current Password field.
   • Enter the desired new Password in the New Password field.
     • The password must follow the Symbient AI Password Policy.
   • Enter the new password again in the New Password Again field.
5. Click Submit when finished.
6. Click the Dismiss button to return to the Details view page.

2-Factor Authentication Tab

If 2-Factor Authentication is enabled, users will be required to enter an authentication code provided by an Authenticator app on their mobile device when they log into Symbient AI. For information about enabling 2-Factor Authentication, please see Enable / Disable Two Factor Authentication.

To set up 2-Factor Authentication:

1. Ensure that you have a valid Authenticator app such as the Google Authenticator or Microsoft Authenticator.
2. Click on the Settings icon on the Menu Bar.
3. Click on the 2FA tab.
4. Set up a connection between Symbient AI and your Authenticator app by scanning the QR code.
   • Alternatively, you can enter the Auth Secret code above the QR into the Authenticator app on your mobile device to also set up a connection.
5. Then, type the code provided by your Authenticator app into the field labeled 2FA Code from Authenticator.
6. Click on the Confirm Code button to verify app authentication.
   • If the authorization fails, a red message will appear warning you that authorization failed. Please check your code or enter it again.

Organization Tab

The Organization tab allows users to define a universal name for their Organization’s Chatbot along with their Organization information.

To configure your Organization, do the following:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Organization tab.
3. These fields are required to save the configuration:
   • Bot Name
   • Organization Name
4. These fields are optional:
   • Address 1
   • Address 2
   • City
   • State (2-letter abbreviation)
   • Zip
   • Country (2-letter abbreviation)
   • Phone
   • Email
   • Industry
5. Click the Save button to save the current Organization details.

Membership Tab

The Membership tab is used to view, change, or cancel your Symbient AI Plan. If a user is currently enrolled in membership, it will be shown next to "Current Plan:".

Selecting or Changing a Membership Plan

To change the plan or to select a plan for the first time, do the following:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Membership tab.
3. Press the Choose Plan button below the desired Plan.
   • To see Monthly vs. Yearly pricing options for different plans, toggle the slide bar in the upper right corner of the window.
4. Once prompted enter:
   • Payment Method
     • If adding a New Payment Method, enter the following:
       • CardHolder Name
       • Card Number
       • Expiry Month
       • Expiry Year
       • Security Code
       • Billing Zip Code
       • Check "Set as Default Payment Method" to make the above payment method the default.
     • If using a Saved Payment Method, select the card to be charged from the Payment Method dropdown menu. For more information about Saved Payment Methods, please see Payment Methods.
   • Check the authorization checkbox above the Submit button. This acknowledges that you authorize Symbient AI to continue charging your payment method either monthly or yearly, based on the plan you have selected.
5. Click Submit.
   • To cancel the payment or select another plan, click Dismiss.

Cancel Membership

To cancel a currently active Membership plan, do the following:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Membership tab.
3. Click the Cancel Membership button to the right of "Current Plan."
4. When prompted to Confirm, Click Yes.
   • To maintain membership, Click No.

Payments Tab

The Payments tab shows any Successful, Scheduled, and Canceled payments. The "Date" column refers to when the payment was initiated. If it is a Scheduled payment, it will show the date the payment will be processed in the "Next Due Date" column.

Payment Methods Tab

The Payment Methods tab is used to save or delete Payment Methods and can mark a Payment Method as the Default Method.

Note: For security reasons, payment methods are NOT stored on Symbient AI servers. They are only stored on Opta Pay, which is a PCI-compliant payment processor of Lucus Labs and Symbient AI.

Add Payment Method

To Save Payment Information:

1. Click on the Settings icon on the Menu Bar.
2. Click on the My Payment Methods tab.
3. In the "Add Payment Method" section to the right of the window, enter the following required information:
   • CardHolder Name
   • Card Number
   • Expiry Month
   • Expiry Year
   • Security Code
   • Billing Zip Code
4. Click Save to finish.

Set Default Payment Method

To set a Payment Method as the Default:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Payment Methods tab.
3. Next to the Payment Method you want to set as the default, click Set as Default.

Delete Payment Method

To Delete Payment Method:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Payment Methods tab.
3. Next to the Payment Method you want to Delete, click Delete.
   • If a Scheduled Payment is associated with the Payment Method, the Payment Method cannot be deleted.
   • If a scheduled Payment needs to be canceled, please Contact Symbient AI Support.

Integrations Tab

To integrate with Symbient AI, you will use the API Key and ENC Key found on the Integrations tab in Account Settings.

Symbient AI will generate these keys for you to use with your integrations.

Reviewing API and ENC Keys

To find the API and ENC Keys:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Integrations tab.
   • Here you will find your "API Key" and "Enc Key".
   • These will be needed for any integrations with Symbient AI.

Note: You can click the Copy icon at the end of each field to copy the value to your clipboard.

Important Security Notice

If you suspect that your API and ENC keys have been compromised, it's important to take immediate action to secure your account and prevent unauthorized use. Please follow these steps:

1. Log in to Your Account: Access your Symbient AI account by logging in through the main website.
2. Revoke or Renew Your Keys: See the Revoking API and ENC Keys or Renewing API and ENC Keys sections below.
3. Notify Us: After securing your account, please Contact Symbient AI Support to inform us of the potential compromise. This is an essential step, as we may need to adjust your API usage or credit back any limits that may have been used by the unauthorized party.

Revoking API and ENC Keys

Revoking API and ENC keys will disable any currently active keys without enabling new keys.

To Revoke API and ENC Keys:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Integrations tab.
3. Click Revoke Keys.
   • When Prompted with the message "This action will cause any applications using these keys to stop working. Are you sure you want to continue?"
     • Click Yes to Revoke Keys
     • Click No to Cancel

Renewing API and ENC Keys

Renewing API and ENC keys will disable any currently active keys AND will enable new keys. These new keys will be visible in the Integrations tab. For more information about finding the new API and ENC keys, please see Reviewing API And ENC Keys.

To renew API and ENC Keys:

1. Click on the Settings icon on the Menu Bar.
2. Click on the Integrations tab.
3. Click Renew Keys.
   • When prompted with the message "This action will cause any applications using these keys to stop working. Are you sure you want to continue?"
     • Click Yes to Revoke and Renew Keys
     • Click No to Cancel

Logout

To log out of Symbient AI from within the Symbient AI application, click the Logout at the bottom of the Menu Bar.

Symbient AI API Documentation

For information about the Symbient AI API, please see https://symbient.ai/page/docs/apis.

Glossary of Terms

Activity

Activities are predefined actions or instructions that can be run within other areas of Symbient AI. An Activity gives a set of instructions to an LLM Model. These instructions can be any natural language directions such as "Summarize this transcript" or "Retrieve all emails mentioned in this text."

Automation

Automations allow you to create in-depth logic systems that can perform a wide variety of functions. These Automations use individual steps called Operators, which can execute tasks such as reading or sending emails, running an Activity, passing text to an LLM Model, or reading and writing data to a webhook. These Operators are chained together to form an Automation Chain.

Menu Bar

When a user is logged in, there will be a menu on the left-hand side of the screen that allows the user to navigate between the main pages of the application.

Roles

• Company Admin: This role has access to do anything in Symbient AI. They can add/edit/delete other users and models, and even maintain their group's membership, payments, and payment methods.
• Accounting Manager: Needed for viewing Payments and Payment Methods tabs on the Settings page.
• Activities Manager: Authorized to access, add, edit, and delete Activities. Other users can use these Activities on Transcriptions, Automations, and other pages, but they are not allowed to modify or delete them.
• Actors Manager: Authorized to access, add, edit, and delete Actors. Other users can use these Actors, but they are not allowed to modify or delete them.
• Automations Manager: Needed for maintaining Automations.
• Model Manager: Authorized to access, add, edit, and delete models. Other users can use these models on the Transcriptions, Automations, and other pages, but they are not allowed to modify or delete them.
• User Manager: Authorized to add, edit, and/or delete other users from the Users page.

Model

A model in the context of Large Language Models (LLMs) is a system trained to process, understand, and generate human language. They function as a way to switch between different use cases and specializations within Symbient AI applications. There are three primary classifications of Symbient AI models, each designed for distinct operational purposes:

LLM (Large Language Model)

An LLM Model can pull information from .txt, .doc, .docx, and .pdf Data Sources and can utilize a specialized Vocabulary for better results. The primary function of an LLM is to interpret user requests and generate responses by leveraging its vast repository of text information.

ASR (Automatic Speech Recognition)

An ASR Model is used when processing Transcriptions. The main function of the ASR Model is to help Symbient recognize specific words or phrases it might not know, such as technical terms or industry-specific language, like medication names or legal jargon.

Classification

The Classification model is used to analyze user requests and identify Intents. It is particularly useful when trying to classify different types of user engagements, such as distinguishing positive from negative feedback, or identifying requests for transactions, such as placing or modifying orders.

AI Actor

An AI Actor refers to a set of additional instructions added to the base model to refine its response mechanism. These instructions are designed to tailor the model's output, ensuring it aligns with specific contextual or tonal requirements. For example, you could instruct it to:

• Be an Expert on a Specific Topic: "Answer all questions as if you're an expert on Symbient Opta and its uses."
• Maintain Specific Documents: "Your role is to ensure the user documentation for my website is always current."
• Adopt a Unique Personality: "Pretend you are a pirate in all your responses."

Intent

Intents are the underlying purpose or objective within a user's query or statement. They are used within Models to train the Slots and Input tabs. Some examples of user intent extracted from a request could be positive vs. negative customer interaction or identifying when a user wants to place or change an order.

Slots

Slots are predefined variables within a conversational model that are used to extract and process specific pieces of information from user input. These variables allow the model to understand and act upon the details of a user's request. For instance, in the context of a customer seeking to book a flight with an airline, the model could have Slots for travel parameters such as the departure city, destination, and travel date.

The data within these Slots can be extracted and transmitted to other systems for storage, verification, and initiating actions, such as triggering a travel-booking system to auto-fill a reservation. This structured approach enables the model to accurately identify and utilize the relevant information provided by the user, thereby facilitating a more efficient and tailored response to the user's needs.