A data fabric is a design framework for creating flexible and reusable data pipelines, services, and semantics. It uses data integration, active metadata, knowledge graphs, profiling, machine learning, and data cataloging. Data Fabric changes the main approach to data management, which is “build to suit” for data and use cases and replaces it with “observe and leverage”.

Data governance is a framework of rules and guidelines for how everyone should handle and use information in a company to keep it accurate, secure, and useful.

Why it matters: Strong governance ensures data is secure, compliant, and used responsibly, reducing risk while enabling business value.

Data integration involves practices, techniques, and tools to ensure consistent access and delivery of data across different areas and structures within a company. It aims to meet the data needs of all applications and business processes efficiently.

Why it matters: Seamless data integration ensures consistency, reduces duplication, and supports accurate, organization-wide insights.

Data lineage refers to the lifecycle and journey of data from origin to destination. From creation to how it’s been edited, transformed and used. Data lineage is critical to know how data can and should be used (compliance), how it was generated and how trustworthy it is. This becomes particularly important when considering data for insights or for machine learning and large language models.

Data management refers to the collection, organization, protection and utilization of an organization’s data and is a core enabler of modern businesses. Data is considered a company’s most critical and valuable asset, however without tooling to effectively manage and make use of the data, it is worthless. Data management technologies include Master Data Management, Customer Data Platforms, Data Unification platforms and Data integration platforms. Every system in use at an enterprise collects data, so a clear data management strategy is crucial to manage, govern and make use of all the data in a safe, secure and compliant way.

Data mesh is a data management approach that supports a domain-led practice for defining, delivering, maintaining, and governing data products. While it’s not yet an established best practice, data mesh helps ensure that data products are easy to find and use by data consumers, such as business users, data analysts, data engineers, or other systems. Additionally, data products must meet terms of service and SLAs, forming a contract between the provider and the consumer.

Data modelling is the practice of designing entities, attributes, and the relationships between them to represent real-world concepts within a system. It determines how information is organized and how different pieces of data connect.

Why it matters:
Clear data models reduce ambiguity in how data is interpreted by both humans and AI. By explicitly defining relationships, data modelling helps AI systems reason more accurately and consistently across complex domains.

Data observability is the ability to see and understand what data systems are doing in real time, including how data changes, moves, and behaves across pipelines. It focuses on detecting issues such as anomalies, drift, delays, or unexpected changes as early as possible.

Why it matters:
Strong data observability helps organizations catch data drift and abnormal behavior before it causes downstream impact. For AI systems, this early visibility is critical to maintaining reliability, accuracy, and trust.

Data poisoning, also known as AI poisoning, involves a deliberate and malicious contamination of data to compromise the performance of AI and ML systems. Attackers may inject false, misleading, or manipulated data into the training process to degrade model accuracy, introduce biases, or cause targeted misbehavior in specific scenarios.

Learn how to protect agaist data poisioning here.

Data profiling involves statistical analysis of various datasets (both structured and unstructured, external and internal) acting as an enabler to provide business users with insights into data quality and identify data quality issues. Profiling also checks data against established rules from rules management.

Data provenance is a historical record of source data, a way to understand the journey of data throughout the organization. Data provenance plays a crucial role in understanding the quality of your data and ensuring its veracity. It’s like a detailed travel log for data, showing where it came from, where it has been, and how it has changed over time.‍

Why it matters: Knowing a data’s origin and transformations ensures accuracy, trust, and compliance across systems and applications.

Data risk scoring is a method of rating potential risk on different kinds of information based on how sensitive it is and how likely it could be accessed by someone who shouldn't have it.

Why it matters: Understanding potential data risks allows organizations to prioritize mitigation and protect sensitive information.