Skip to main content
This glossary provides definitions for key terms, concepts, and technologies as they relate to the Ekyam Solution. Understanding these terms will help in better utilizing the platform’s features and functionalities. This glossary covers many of the key terms from our document. As we continue to expand our sections, more specific AI and developer-related terms can be added. 

​
A

  • Agentic AI: The Ekyam solution layer utilizes AI, large language models (LLMs), and agent-based architectures to enable smart interactions, generate insights, and support autonomous operations for retailers.
  • AI Agents: Specialized AI agents are designed to carry out tasks, support users, or make decisions within specific retail domains. 
  • AI-Powered Data Mapping: Ekyam uses Artificial Intelligence to analyze source and target data schemas (including EDI/iDOC structures) and suggest field mappings, simplifying integration setup.
  • Actions (Ekyam Workflows): It includes specific actions performed within a workflow such as API calls, database lookups, EDI/iDOC processing, or sending notifications. 
  • Apache Kafka: An open-source, distributed event streaming platform used by Ekyam as its central message backbone for high-throughput, fault-tolerant, and scalable data handling.
  • Asynchronous AI Task Management: The use of Apache Kafka within Ekyam to queue and manage AI tasks that are long-running, decoupling them from immediate user interaction.
  • API (Application Programming Interface): A set of rules and protocols that allows different software applications to communicate and exchange data. Ekyam’s Universal Connector supports various API standards.
  • API Key Authentication: An authentication method where an API key (a unique string) is used to grant access to an API. Supported by Ekyam’s Universal Connector.
  • Authentication: The process of verifying the identity of a user, system, or application attempting to access Ekyam. Ekyam primarily uses JWT (JSON Web token) for this.
  • Authorization: The process of determining what actions an authenticated user, system, or application is permitted to perform within Ekyam, often managed by Role-Based Access Control (RBAC).
  • Authorization Code Grant (OAuth 2.0): An OAuth 2.0 flow used by web applications where a user grants permission for an application to access their resources.

​
B

  • B2B Document Exchange: The process of businesses electronically exchanging structured documents like purchase orders, invoices, and shipping notices, often using EDI or iDOC formats. Ekyam facilitates this.
  • Batch Jobs: A traditional integration method involving periodic extraction, transformation, and loading of data between systems at scheduled intervals. Ekyam aims to reduce reliance on these.
  • Basic Authentication (HTTP): A simple authentication scheme built into the HTTP protocol, where the client sends a username and password with each request. Supported by Ekyam’s Universal Connector.

​
C

  • Chronicle: Ekyam’s unified data engine that stores and contextualizes all product and inventory activity across the retail stack. 
  • Canonical Data Models (Ekyam Data Standards): Standardized representations for key business entities (e.g., Product, Order, Customer, Inventory) are defined by Ekyam to ensure consistent data interpretation.
  • CDC (Change Data Capture): A technique for tracking and capturing data changes in a database by reading its transaction logs, enabling near real-time event detection for Ekyam Event Listeners.
  • Chunking (RAG): The process of breaking down large documents into smaller, semantically coherent pieces for optimizing context retrieval for Ekyam’s Retrieval-Augmented System (RAG).
  • Conditional Logic (Ekyam Workflows): The ability within Ekyam Workflows to execute different branches of steps based on specified conditions (e.g., IF/THEN/ELSE, Switch/Case).
  • CRM (Customer Relationship Management): Systems used to manage customer interactions, data, and relationships. Ekyam integrates with CRMs.
  • Configuration Isolation (Multi-Tenancy): An architectural principle in Ekyam ensuring that each client’s specific configurations (e.g., connectors, workflows, EDI/iDOC profiles) are separate and do not affect other clients.
  • Confidence Scoring (AI Data Mapping): A score provided by Ekyam’s AI with each mapping suggestion, indicating the AI’s certainty about the match. 

​
D

  • Data Isolation (Multi-Tenancy): A core security principle in Ekyam ensuring that each client’s data is strictly separated and inaccessible to other clients.
  • Data Warehouses: Ekyam integrates with the centralized repositories for storing large volumes of raw and processed data for analytics. 
  • Data Mapping: Ekyam does AI-powered data mapping. It is the process of defining correspondences between data fields from a source system and a target system. 
  • Data Silos: A situation where information is isolated within individual systems, preventing a holistic view of business operations. Ekyam aims to eliminate these.
  • Document Loaders (LangChain): LangChain components used to ingest data from various sources (files, web pages) for processing in RAG pipelines.
  • Decoupling (EDA): An architectural benefit of Event-Driven Architecture where event producers and consumers operate independently, communicating indirectly, which enhances resilience and scalability. Ekyam leverages this.

​
E

  • eCommerce Platforms: Eykam integrates with the software applications that power online stores. 
  • EDA (Event-Driven Architecture): It is an architectural paradigm where system behavior is orchestrated by the production, detection, and consumption of events. This is a core principle of Ekyam.
  • EDI (Electronic Data Interchange): A set of standards for structuring information to be electronically exchanged between businesses. Ekyam provides robust support for parsing, processing and generating EDI documents. 
  • Embedding Models (RAG): Machine learning models used in Ekyam’s RAG system to convert text or data chunks into numerical vector embeddings that capture semantic meaning.
  • ERP (Enterprise Resource Planning): Integrated management software for core business processes. Ekyam integrates with ERPs, including SAP systems via iDOCs.
  • Event Listeners (Ekyam): Components of Ekyam that actively monitor connected source systems (via polling, webhooks, CDC, file detection) for business events, acting as the platform’s sensory network.
  • Event-Based Triggers (Ekyam Workflows): Workflow initiators that automatically start a workflow when a specific event is detected by an Ekyam Event Listener.

​
F

  • FTP/SFTP Servers: File Transfer Protocol / Secure File Transfer Protocol servers used for file exchange. Ekyam’s Universal Connector can connect to these, often for EDI/iDOC transfer.
  • File-Based Listeners (Ekyam): Event listeners that monitor specific directories (e.g., on FTP/SFTP, cloud storage) for new or modified files, including EDI or iDOC documents.
  • Fine-Tuning (LLM): The process of further training a pre-trained LLM on a smaller, domain-specific dataset to adapt its knowledge and improve its performance on particular tasks.
  • Functional Microservices (Ekyam): Core Ekyam platform capabilities (e.g., authentication, order processing, inventory ledger) designed as independent, scalable microservices.

​
G

  • Graphical Workflow Designer (Ekyam): An intuitive, visual drag-and-drop interface within Ekyam for designing and configuring business process workflows.

​
H

  • Hallucinations (LLM): A phenomenon where LLMs generate plausible but incorrect or fabricated information. Ekyam’s RAG architecture helps mitigate this.

​
I

  • iDOC (Intermediate Document): A standard data container format used by SAP systems for exchanging business transaction data (e.g., orders, deliveries, invoices) with other SAP or non-SAP systems.
  • iDOC Parsing Engine: A component within Ekyam’s Universal Reader responsible for interpreting the structure and content of incoming SAP iDOC files. 

​
J

  • JSON (JavaScript Object Notation): A lightweight data-interchange format commonly used in REST APIs. Ekyam processes and generates JSON.
  • JWT (JSON Web Token): An open standard for securely transmitting information between parties as a JSON object, used by Ekyam for authentication and authorization.

​
L

  • LangChain: An open-source framework used by Ekyam for developing applications powered by LLMs, providing modular components for chains, agents, memory, and tool usage.
  • LangGraph: A library built on LangChain, used by Ekyam to create stateful, multi-actor AI applications with LLMs by modeling them as cyclical graphs.
  • LangSmith: A platform for debugging, testing, evaluating, and monitoring LLM applications, used by Ekyam to ensure the observability and reliability of its LangChain-based AI agents. 
  • Large Language Models (LLMs): Advanced AI models (e.g., OpenAI, Gemini) capable of understanding and generating human-like text, forming a core part of Ekyam’s Agentic AI.
  • Logical Data Segregation: A primary method for data isolation using a client’s ID to ensure all operations are confined to the data of the authenticated client. 

​
M

  • Machine Learning (ML): A field of AI that enables systems to learn from data without being explicitly programmed. Ekyam uses ML for AI-powered data mapping and potentially in its analytical models.
  • MCP (Model Context Protocol): A conceptual framework inspiring Ekyam’s approach to structuring how AI agents access and utilize context, tools, and resources for effective task performance.
  • Microservice Architecture: Ekyam’s platform-wide design where core functionalities and connectors are built as small, independent, and scalable services that communicate with each other.
  • Middleware: Software that acts as a bridge between other applications, databases, and services, facilitating communication and data exchange. Ekyam functions as an advanced AI-powered middleware for retail.
  • Multi-Tenancy: An architecture where a single instance of a software application (Ekyam) serves multiple clients (tenants) while keeping their data and configurations isolated and secure.

​
N

  • Nodes: Units of computation within a LangGraph, representing LLM calls, tool invocations, or custom functions, used to build Ekyam’s AI agents.
  • Natural Language Processing (NLP): A field of AI that enables computers to understand, interpret, and generate human language. Used by Ekyam’s AI for conversational interfaces and data mapping.
  • NoSQL Databases: Databases that do not use the traditional relational (tabular) model, such as document stores (MongoDB) or key-value stores (Redis). Ekyam can connect to these. 

​
O

  • OAuth 1.0/1.0a & OAuth 2.0: Authorization frameworks that allow third-party applications to access user resources without exposing credentials. Supported by Ekyam’s Universal Connector.
  • OMS (Order Management System): Software that manages the entire order lifecycle. Ekyam integrates with OMSs.

​
P

  • PIM (Product Information Management): Systems for centralizing and managing product information. Ekyam integrates with PIMs.
  • Pinecone Vector DB: A managed vector database used by Ekyam’s RAG system for efficient semantic search and retrieval of contextual information to ground LLM responses.
  • POS (Point of Sale): Systems used in physical stores to process transactions. Ekyam integrates with POS systems.
  • Polling (Event Listeners): A method where Ekyam Event Listeners periodically query source systems to check for new or updated data.
  • Proactive Agents (Ekyam AI): AI agents in Ekyam that can monitor events and data, identify situations requiring attention, and initiate actions or alerts without direct user command.
  • Prompts (MCP/LLM): Carefully engineered instructions given to LLMs to guide their behavior, reasoning, and response generation within Ekyam’s AI.
  • Protocol Handling (Universal Writer): The capability of Ekyam’s Universal Writer to use the appropriate communication protocol (e.g., HTTP, SFTP) to deliver data to destination systems.

​
Q

  • Quantity Committed: Inventory reserved for sales orders but not yet shipped. Tracked by Ekyam Universal Ledger.
  • Quantity OnHand (QOH): Total physical stock at a location. Tracked by Ekyam Universal Ledger.
  • Quantity OnOrder: Stock ordered from suppliers but not yet received. Tracked by Ekyam Universal Ledger.

​
R

  • RAG (Retrieval Augmented Generation): An AI architecture used by Ekyam that enhances LLM responses by first retrieving relevant factual context from external knowledge sources (like Pinecone) and providing it to the LLM.
  • RBAC (Role-Based Access Control): A security model used by Ekyam to manage user permissions based on assigned roles within a tenancy.
  • Redis: An in-memory data store used by Ekyam for high-performance caching in its AI systems, reducing latency and costs.
  • Refresh Token Grant (OAuth 2.0): An OAuth 2.0 mechanism to obtain new access tokens without requiring user re-authentication.
  • Resilience (Architecture): The ability of the Ekyam platform and its microservices to withstand and recover from failures, ensuring high availability.
  • Resources (MCP): Data and knowledge assets (e.g., Universal Ledger data, EDI/iDOC content, product catalogs) that Ekyam AI agents can access and reason over.
  • REST (Representational State Transfer) APIs: A common architectural style for web APIs, widely supported by Ekyam’s Universal Connector.
  • Retry Mechanisms (Kafka/Workflows): Ekyam’s capability to automatically retry failed operations (e.g., API calls, event processing) a configured number of times to handle transient issues.
  • RouterChain (LangChain): A LangChain component used by Ekyam’s intelligent LLM router to decide which LLM or prompt to use for a given query.

​
S

  • Safety Stock: A buffer inventory level maintained to prevent stockouts. Considered by Ekyam’s Universal Ledger.
  • Scalability (Architecture): The ability of the Ekyam platform and its components (like Kafka and microservices) to handle increasing volumes of data and transactions by adding resources.
  • Schema Analysis (AI Data Mapping): The process by which Ekyam’s AI ingests and understands the structure of source and target data to suggest mappings.
  • SCM (Supply Chain Management) Systems: Software used to manage the end-to-end flow of goods, information, and finances in a supply chain.
  • SDKs (Software Development Kits - Ekyam): Libraries and tools provided by Ekyam (initially in Python) to enable developers to build custom extensions and integrations for the platform.
  • Semantic Search (RAG/Pinecone): Searching for information based on meaning and context rather than just keywords, enabled by vector embeddings and Pinecone in Ekyam’s RAG.
  • SequentialChain (LangChain): A LangChain component for linking multiple chains or calls in a sequence, where the output of one becomes the input to the next.
  • Single Source of Truth (SSoT): A central, reliable, and consistent repository of data (like Ekyam’s Universal Ledger) that all connected systems can trust and use for decision-making.
  • System Prompts (MCP/LLM): High-level instructions defining an Ekyam AI agent’s persona, objectives, and constraints.

​
T

  • Tenant ID (Multi-Tenancy): A unique identifier used in Ekyam to associate all data and configurations with a specific client, ensuring logical data segregation.
  • Text Splitters (LangChain): LangChain components used to break down large documents into smaller chunks for effective processing in RAG pipelines.
  • Tools (MCP/LangChain): Capabilities or actions (e.g., API calls, database queries, analytical model invocations) that Ekyam AI agents can perform to interact with their environment.
  • Triggers (Ekyam Workflows): Mechanisms (event-based, scheduled, manual, API) that initiate the execution of an Ekyam Workflow.

​
U

  • Universal Connector : A versatile Ekyam component, built as microservices, that enables connections to a wide array of external systems, APIs, and data sources using various protocols and authentication methods.
  • Universal Ledger: A centralized, real-time, standardized data store within Ekyam that acts as the definitive source of truth for key operational data like inventory, orders, and customer information.
  • Universal Reader: An Ekyam component responsible for ingesting raw data from various source systems (including parsing EDI/iDOCs) and transforming it into Ekyam Data Standards.
  • Universal Writer: An Ekyam component that takes standardized data from within Ekyam and transforms/formats it for delivery to destination systems or trading partners, including generating EDI/iDOC documents.

​
V

  • Vector Embeddings (RAG): Dense numerical representations of text or data that capture semantic meaning, used by Ekyam’s RAG system with Pinecone for similarity searches.
  • Vector Store (Pinecone): A specialized database, like Pinecone, used by Ekyam to store and efficiently query vector embeddings for its RAG system.

​
W

  • Webhooks (HTTP Callbacks): A method where a source system actively sends an HTTP notification to an Ekyam endpoint when an event occurs, enabling real-time data capture.
  • WMS (Warehouse Management System): Software for controlling and optimizing warehouse operations. Ekyam integrates with WMSs.
  • Workflows (Ekyam): The Ekyam component for defining, orchestrating, and automating multi-step business processes that span across integrated systems, incorporating transformations, business logic, and AI capabilities.

​
X

  • XML (Extensible Markup Language): A markup language for encoding documents in a format that is both human-readable and machine-readable. Handled by Ekyam’s Universal Connector, Reader, and Writer.

​
Z

Zep: A platform for persistent, long-term memory for LLM applications, used by Ekyam to enhance the conversational coherence and context-awareness of its AI agents.