Workflow - AXOMA

The Workflow Module in the Axoma Platform enables users to visually design, configure, and manage end-to-end AI-driven automation processes. It acts as the orchestration layer, bringing together AI agents, LLMs, and external tool integrations (e.g., Jira, Salesforce, Gmail, ServiceNow) within a unified, drag-and-drop environment. Workflows can automate multi-step tasks, route information intelligently, and implement dynamic logic to solve complex business scenarios. Axoma

Workflow Management Dashboard

The Workflow Dashboard serves as the central hub for managing all existing workflows. Users can create, edit, publish, or monitor automation processes from one interface.

Feature	Description
Workflow List	Displays all available workflows with key details including Name, Description, Created Date, Status (Active/Inactive), Publish State (Draft/Published), and Workflow Type (Chat, LCNC, etc.).
Create New Workflow	Opens a new workflow design canvas for building automation from scratch.
Search & Filter	Allows searching workflows by name or filtering by status, publish state, or creation date.
Actions	Quick access options such as Edit, Delete, View Runs, Schedule Flow, Open in Playground, and Publish/Unpublish.

Create a Workflow

Click Create New Workflow to enter the Workflow Playground a visual design environment where automation logic is built using pre-defined components.

1. Open the Playground

Click Create New Workflow or select an existing workflow to modify.
The visual canvas opens, displaying a Components Panel on the left and a Design Canvas in the center.

2. Drag Components

From the left-hand Components Panel, drag and drop desired components (e.g., Chat Input, If-Else, DocuChat, Run Agent, etc.) onto the canvas.

3. Connect Components

Use arrows to connect output fields of one component to input fields of another.
The flow defines the execution order and data mapping between steps.

4. Configure Each Component

Click on a component to open its configuration panel.
Fill required fields such as API Key, Model Selection, Query, or Condition Parameters.

5. Test & Publish

Use the Preview/Run feature to test your workflow.
Click Publish when ready to make it active.

Workflow Components

Each component in the Components Panel plays a specific role. They are grouped into functional categories for easy navigation.

Component Categories Overview

Category	Description
Agent	Manages AI agent execution inside workflows.
Chat	Manages conversational flow between user input, LLM response, and chat output.
Control	Adds logical flow control like loops or sequence continuations.
Gmail	Handles Gmail-related operations for reading, replying, or sending emails.
Helper	Provides small but essential utilities like generating IDs or fetching timestamps.
Jira	Automates Jira ticket management and issue tracking.
LLM	Handles tasks related to language models, such as creating embeddings or model calls.
Logic	Implements conditional flow using If-Else branches.
Processing	Data manipulation and transformation utilities.
Salesforce	Integrates Salesforce cases for CRM automation.
Scope	Defines data scope such as chat or app-level context.
ServiceNow	Automates incident management and ticket updates.
Variables	Stores and reuses static or dynamic input values.

Component Details

Agent

Component	Description	Key Configurations
Run Agent	Executes a predefined agent created in Agent Studio.	Select Agent Name, LLM Model, provide API Key, and configure Timeout if needed.

Chat

Component	Purpose	Description
Chat Input	Entry point	Captures user queries or inputs to initiate workflow execution.
Chat Output	Exit point	Displays or returns the final AI response to the user.
DocuChat	Knowledge-based chat	Integrates document-based chat powered by a connected knowledge base. Includes settings for Similarity Score, Embedding Model, and Knowledge Base Selection.
LLM Chat	Model interaction	Routes input through a selected LLM model. You can define System Prompt, User Query, API Key, and Model Type.

Scope

Component	Description
App Scope	Defines global data scope for app-level variables.
Chat Scope	Manages conversational history and session-level variables.

Control

Component	Purpose
Loop	Executes a defined set of actions repeatedly based on a condition.
Next (Continue)	Defines sequential flow to continue execution after a previous step.

Helper

Component	Description
Current Datetime	Fetches current date and time dynamically.
ID Generator	Generates unique identifiers for records, messages, or transactions.

LLM

Component	Description
Create Embedding	Generates embeddings for text data to use in retrieval or semantic search workflows.

Logic

Component	Description
If-Else	Adds conditional branching logic. For example: If the LLM output = “Raise Ticket”, then route to Jira Create Issue; else continue chat.

Processing

Component	Description
Merge Data	Combines data from multiple sources.
Combine Text	Merges multiple text fields into one.
Data to DataFrame	Converts structured data into a dataframe format for advanced manipulation.
RegEx Extractor	Extracts specific patterns using regular expressions.

Gmail

Component	Purpose
Draft Gmail	Creates a draft email within a connected Gmail account.
Gmail Extractor	Extracts email metadata such as sender, subject, or date.
Read Gmail	Reads the contents of emails from the inbox based on filters.
Reply Gmail	Sends a reply to a specific email thread.
Send Gmail	Sends new emails to specified recipients.

Jira

Component	Description
Create Jira Issue	Creates a new issue/ticket in Jira.
Jira Ticket Extractor	Extracts relevant details from Jira tickets.
Jira Read Issues	Fetches issue data from Jira based on filters.
Jira Update Issues	Updates existing Jira tickets (status, description, etc.).

SalesForce

Component	Description
Create Case	Creates a new case in Salesforce.
Get Multiple Cases	Retrieves a list of cases.
Get Single Case	Fetches details of a specific case.

ServiceNow

Component	Description
App Scope	Defines global data scope for app-level variables.
Chat Scope	Manages conversational history and session-level variables.

Variable

Component	Description
Static Input	Defines fixed input variables (constants) for reuse across the workflow.

Workflow Execution Flow

A workflow executes step by step, following the defined connections:

Input: Captured via Chat Input or API trigger.
Processing: Handled by LLMs, Agents, or external integrations.
Logic Handling: Decisions made by If-Else or Loop components.
Output: Final message displayed via Chat Output or logged in connected systems.

Best Practices

Keep workflows modular reuse existing templates for common automations.
Validate each node before publishing.
Use naming conventions (e.g., “WF_SupportTicket”) for clarity.
Regularly review published workflows for optimization.

​Workflow Management Dashboard

​Create a Workflow

​1. Open the Playground

​2. Drag Components

​3. Connect Components

​4. Configure Each Component

​5. Test & Publish

​Workflow Components

​Component Categories Overview

​Component Details

​Workflow Execution Flow

​Best Practices