Comment on page
Check is an AI action to set up interaction to get a specific value or match with a term or a statement. You can prompt a customer, get a webhook response from your back-end system, or both.
You can set up how the Subject will terminate. In closing, you can turn the Restart and Ending messages on and off, and customize the messages as you need.
Each Component in a Subject can have its own Completion, which determines the completion of flow building. Completion actions include Show Response, Confirm, Solve, and Validate Result; they are conditional on the Subject types and Components.
Components define conversation structures in a Subject. There are two types of Components: Normal Components and Diagnostic Components. Normal Components include Selection Classes, Selections, Conditions, Parameter Bundles, and Parameters. Diagnostic Components include Diagnostic Selection Classes, Diagnostic Selections, and Diagnostic Conditions. Diagnostic Components are only available for Diagnosis Subjects since they are for diagnosing problems.
When you audit AI action details in Sandbox or Conversation history, you can see where the Component is located in the Subject in Component Structure. It shows the Step where the Component is, and its parent Conditions as well.
Compound Value Parameter is used whenever a user wants to let their customer put multiple values that will be recognized by our Mind Expression. For example, getting “List” as data type from JSON response.
Condition is a Component that is a dependent condition on the Subject. Conditions are very similar to Subjects, but they do not stand alone; they exist as a part of a Subject. Conditions are available in Query Freestyle and Diagnosis Subjects.
Confirm is an AI action to set up a confirmation. Confirm fetches a binary value: either an affirmation or a negation. Upon affirmation, it continues to the next step in the conversation. If negation is fetched, it can either try again, which means that the Condition or Selection in which it exists is re-called, or it can terminate, which means the session comes to an end. You can prompt a customer, get a webhook response from your back-end system, or both.
You can audit every AI interaction in Conversation History. You can see how the AI is interacting with your customers and the Sandbox test sessions. On the session details page, you can see the AI action details of every AI message.
This is the main parameter to get the data from Webhook.
This is the JSON Path where Mind Expression gets the values from.
Desired Result is a diagnostic goal of a Diagnosis Subject, which is the resolution of the Problem Statement.
The Diagnosis Subject assists in diagnosing problems and providing solutions to the customers by handling subordinate issues to resolve the main problem. The Subject can be used when there are parts to check which can determine what and where the problem lies, and in diagnosing the exact cause of the disruption, to solve the problem as a whole. The goal of the Diagnosis Subject is to guide the customer to the desired result of the Subject.
A Diagnostic Condition is a Component which is a possible cause of the Subject problem. Each Diagnostic is also a potential problem to solve. To resolve the problem that your customer is experiencing, the AI will examine Diagnostic Conditions in the Subject.
A Diagnostic Selection is a Component that defines a choice. Diagnostic Selections belong to a Diagnostic Selection Class.
A Diagnostic Selection Class is a Component that creates Diagnostic Selections to define a choice. By having a Diagnostic Selection Class, you can set up diverging paths (Diagnostic Selections) for your customers to follow in the conversation. A Diagnostic Selection Class holds multiple Diagnostic Selections. To resolve the problem that your customer is experiencing, the AI will examine Diagnostic Selections in the Subject, which are the possible causes of the problem.
An Ending message is a message that is displayed upon session termination. You can set up the message in Closing on the Subject building page.
Define the type of Parameter value to receive from the customer or system.
All entities defined by users are listed in the My Entities tab. Users can add and manage user-defined entities here.
Using an Escalation URL, which is an optional webhook, you can send the entire chat script to your enterprise backend system when an escalation is performed in case of unsuccessful session termination.
Fallback is a situation the AI couldn't successfully get a response to the customer message. You can audit the sessions with fallback in Conversation History.
The Mind Expression reasoning engine continuously learns and adapts new ontologies. In Knowledge, you can also educate the AI with your domain knowledge. At the moment, we support managing entity types and entities.
Literal Entities are the values that are also known as a literal value, which is the value used by API interaction. You can manage the literal entities in Knowledge.
Literal Entity Block consists of the collection of Literal Values variations and Entities.
Literal Value is the value used by API interaction for literal entities. You can manage the Literal Values in Knowledge.
A logicless template for creating dynamic contents. Mustache Template Language will be used further for Parameter, Parameter Bundles, and Result Mapping
A network of utterances in a given subject area or domain. The classification is based on semantic relations such as hypernym, hyponym, holonym, and meronym.
Parameters are the values that will be stored inside Mind Expression, for other uses such as getting the Name of the end-user, and that name will be used in the Conversation inside the Subject.
A Parameter Bundle is a Component that holds one or more Parameters. By having a Parameter bundle, you can set up Parameters. Parameters are values that must be provided to obtain the result for the query. They can be set as required or optional, and the AI will only check required ones with the customer or system to get the Parameter value. Optional Parameters are not invoked unless the customer brings them up in the context.
A Problem Statement is the expected statement of the problem from the customer.
Problem Term is the literal value to match that directs the conversation to the Diagnostic Condition when it is called out.
The Q&A Subject assists in providing a pre-defined answer to a question. It is suitable for simple Q&As that consist of a pair of a question and an answer. It helps you answer questions a customer has, or to find specific information from your database. You can give pre-written answers or have a webhook to find an individual's personal data.
The Query Freestyle Subject helps you take care of customer queries that are more complex to process than simple information retrieval or Q&A handling. It takes care of particular customer requests, it answers questions that a customer might have or finds specific information from your database. You can give pre-written answers or have a webhook to find an individual's personal data.
A Query Statement is the meaning of the expected query from the customer for activating the condition.
Query Term is the literal value to match that directs the conversation to the Condition when it is called out.
This is the expected statement of the problem from the customer. This is available for Diagnosis Subject Type.
This is the expected query from the customer. This is only available for Subject Types: Q & A, Info Search, and Query Freestyle.
A reference creates a shift in the current conversation flow. Having a reference means the conversation will be directed to the selected Component. Diagnostic Components (Diagnostic Selections and Diagnostic Conditions) can be a reference for both Diagnostic Components and normal Components (Selections and Conditions), but normal Components only can be a reference for normal ones (Selections and Conditions).
Regex Entities are the entities with Regular Expressions which are sequences of characters that specify a search pattern. Regex entities are useful for string-searching algorithms or input validation.
A regex pattern is a search pattern in Regular Expression format which are sequences of characters that specify a search pattern.
Regex Pattern Block consists of a set of search patterns that defines the search algorithm and its description.
A Restart message is a message that is displayed to check for a session restart before ending the session. You can set up the message in Closing on the Subject building page.
These are the data that will be extracted from the webhook response. The result mapping will be stored inside the Scope, and can be used within the Conversation.
While building your conversational AI with Subject, Sandbox allows you to test your AI as if your customers were interacting with it. It uses the same API internally, so all the features that are available for API users will be shared. Every interaction in Sandbox will also be logged in Conversation History as well as live sessions so that you can audit your test sessions anytime.
In Mind Expression, Scopes work as independent AI agents. A Scope is a minimum service module in which configurations of endpoint APIs and sets of Subjects are defined. You can create one or more Scopes, and you will see a list of managed Scopes and get their details from the list. When creating a Scope, the Engine URL (API Endpoint) and Auth Key (authentication key) are auto-generated, which are inherent to the Scope.
A Selection is a Component that defines a choice. Selections belong to a Selection Class.
A Selection Class is a Component that creates Selections to define a choice. By having a Selection Class, you can set up Selections to define diverging paths for your customers to follow in the conversation. A Selection Class must include multiple Selections.
Selection Term is the literal value to match that directs the conversation to the Selection when it is called out.
Show Response is an AI action which you can display a response to your customer. To provide a response to your customer, you can manually set up the response and display it, or you can collect the response from a webhook and show it to your customer.
This is a value coming from the Webhook Response Body, in JSON format. The value is assigned to its label, allowing to be utilized for Conversational purposes.
Solve is an AI action which you can provide solution to solve the problem your customer is facing, and optionally verify the solution. To provide solutions to your customer, you can manually set up the solution and display it or you can collect the response from webhook and show it to your customer. You can solve the problem with customer, or with webhook, or both.
Steps are useful to manage the conversation flow in a Subject when it gets more complex, for example by having multiple Components. It helps you organize the flow better and easily look at the whole picture. Every step is required to include one Component, and it can have child Components if necessary. To add a new sibling Component, you need to add another Step.
As real conversations always entail subjects to cover, Mind Expression enables human-like communication with Subjects. A Subject is what matters to the customer consisting of Components and their associated actions as well as a response message. 4 types of Subjects help you achieve different session goals: Q&A, Info Search, Query Freestyle, and Diagnosis.
System entities are the default system entities available for you without any initial setup. There are system entities that are extensible and inextensible. For extensible system entities, you cannot edit or delete them, but extend them as you need. All inextensible system entities are available for view only. These entities cannot be edited, deleted, or extended.
User defined entities are the entities created and managed by you. All entities defined by you are listed in My Entities tab in Knowledge.
Validate Result is an AI action in which you can validate the desired result of the Diagnosis Subject. This action must be set up for a Diagnosis Subject before publishing it, and it can also be added to a Diagnostic Component as flow completion. To provide solutions to your customer, you can manually set up the solution and display it or you can collect the response from webhook and show it to your customer.
The way your AI communicates with external enterprise systems is by using webhooks. A webhook is a way for a system or app to provide another system or platform with real-time data in JSON, XML, or form-encoded serialization. Mind Expression is currently only allowing JSON as a standard data interchange format. The main purpose of a webhook is for data integration between our platform and the enterprise’s backend system using a service URL.