Key Concepts

On this page, we will explain the key concepts to help you better understand and start using GoInsight.AI.

1. Basic Concepts

1.1 Workflow

A workflow refers to the process of automating a series of tasks with minimal human intervention, typically using software tools.

It's designed to increase efficiency, reduce errors, and streamline repetitive tasks in various business functions (such as HR, finance, or customer service).

Workflow automation isn't a new term. It's been around for decades, but it's now evolving with AI. AI workflows are more intelligent, adaptable and excel in complex environments.

Here are the key differences between traditional workflow and AI workflow:

In short, AI workflow incorporates AI technologies that enable the system to learn from data and adapt dynamically. It can handle more complex and variable tasks without constant human intervention. Traditional workflow, without implementing AI, follows a strict sequence of steps and rules defined by humans. It's predictable and reliable for routine and simple tasks.

In GoInsight.AI, there are two types of AI workflow:

1. Interactive Flow: It's a workflow with a conversational interface that allows users to enter questions and sentences.

2. Service Flow: It's an interface-free workflow to execute background tasks with APIs or triggers.

To learn more about the workflow types, feel free to check the Types of Workflow page here.

1.2 LLM

LLM (Large Language Model) is a type of artificial intelligence model, which is built using deep learning techniques and trained on massive datasets consisting of text and code from diverse sources.

LLMs can understand patterns, context, and semantics in language, then process and generate human language text. Well-known LLM examples include OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude.

So, what can LLMs do in an AI workflow? LLMs serve as powerful components that automate and enhance complex tasks involving language understanding and generation to build intelligent workflows, such as:

• Natural Language Understanding:
  Analyze messages, documents, and emails, understand customer queries, and extract entities, sentiment, and intent.
• Knowledge Management:
  Search and retrieve relevant information from internal documentation and answer questions based on internal data.
• Answer User Queries in Natural Language:
  Provide conversational context and maintain conversational memory.
• Generate Content:
  Generate content like emails, production descriptions, blog posts, code snippets, and marketing copy, or translate content based on prompts.
• Data Transformation:
  Convert unstructured text to structured formats (e.g., JSON, Markdown), and categorize content automatically.

1.3 Nodes

Nodes are the fundamental component of a workflow, representing an individual step or action within the workflow. Connect each node together in a sequence to define the overall flow and logic of the process.

There are multiple basic nodes in GoInsight.AI's workflow, such as LLM, Knowledge Base Retrieval, HTTP Request, and Code. Additionally, a published workflow or a plugin can be used as a node.

To further learn about nodes, feel free to check the Nodes part.

1.4 Connector

A connector is an element that links two or more nodes together, enabling the flow of data or control between them. Connectors define the sequence in which nodes interact, ensuring that data or actions are passed from one node to the next, in the right order you prefer.

When connecting two nodes in a workflow, press and hold the + button on the previous node, then drag the connection line to the next node.

1.5 Canvas

Canvas is the space where you can edit an AI workflow, drag and drop nodes and connectors, and link them in the order that fits your logic.

You can also add notes, and import/export the file on the canvas by right clicking the empty space on the canvas.

You are allowed to zoom in and out of the canvas by simultaneously pressing the Ctrl key and scrolling the mouse.

1.6 Trigger

A trigger in GoInsight.AI workflows is the specific condition or event that initiates the execution of that workflow. It's considered the starting point that tells the workflow when to activate the automated sequence of tasks.

Common types of triggers in AI workflows include:

• User's entering queries
• New data received
• User's submitting a form
• Email received
• Scheduled time
• Webhook or API call

In the Interactive Flow, the user's entering queries on the conversational interface is the trigger that starts the workflow; in the Service Flow, the trigger usually refers to the preset event or scheduled time.

Once you publish a Service Flow, you can set up the trigger easily by clicking the Trigger button under "Save and Publish".

1.7 Prompt

A prompt is an input or instruction given to the LLM to elicit a specific response or output accurately. It serves as the bridge between human intent and AI, guiding the AI on what tasks to perform and how to respond appropriately.

Specificity and clarity are crucial in an AI prompt, as they directly impact how well the AI comprehends and performs the task. The more detailed and precise the instructions, the better outcomes generated by the AI prompt.

If you struggle to write a prompt, don't worry. GoInsight.AI offers an "AI Writing" assistant to help you with that. Simply click the "AI Writing" button, tell the assistant what you want to achieve, then it will generate a prompt for you.

1.8 Knowledge Base

A knowledge base is a repository of information, data, and documentation that AI can reference to make informed decisions, generate responses, or provide solutions, in a more accurate, relevant, and contextual way.

Knowledge bases are essential in workflows involving customer support, recommendation systems, and virtual assistants in industries like e-commerce and retail, healthcare systems, financial services, human resources and talent management, and more.

To learn more about knowledge base, please go to the Knowledge Base page for more details.

2. Technical Concepts

2.1 Variables

Variables are used to store, manipulate, and pass data throughout the various stages of the workflow. They act as placeholders for data and can be dynamically updated based on inputs, outputs, or other actions during the workflow execution.

The variables in GoInsight.AI can be categorized into three different types based on their scope and purpose: System Variables, Conversation Variables, and Internal Variables.

System Variables

System variables refer to system-level parameters preset within the workflow, which can be accessed by other nodes.

GoInsight.AI offers the system variables for Interactive Flow:

GoInsight.AI offers the system variables for Service Flow:

Conversation Variables

Conversation variables are used to store and pass data during each execution of the workflow. Each workflow session has its own independent session variables, so session variables from different workflow sessions do not affect each other. In the conversational workflows, all custom variables are session variables.

Note: This is only applicable to Interactive Flow.

Internal Variables

When a workflow is triggered externally, private (Protected) variables will be restricted, and they will not be exposed or passed in.

2.2 Integration

Integration refers to the process of connecting the AI workflow to existing tools, systems, data sources, and applications so that they can work together seamlessly within the AI-powered process. Integration is a crucial element of AI workflows, as it ensures that the AI system can access external data, trigger actions, and interact with other systems or services as part of a broader business or operational process.

For instance, you can connect your workflow with CRM systems like Salesforce or HubSpot to collect customer data and trigger actions such as sending personalized email campaigns.