DAMA SA Data Management

AI is fundamentally changing how we develop and interact with our data, but throwing a
generic Large Language Model at your database isn't enough. Without proper context, AI
can provide generic answers, hallucinate, or even expose confidential data.

In our upcoming webinar this Thursday, we are diving deep into Designing a Semantic
Layer for AI Model Readiness.

We'll be walking through the exact framework needed to
prepare your models for Copilot, Data Agents, and external LLMs.

Here is a sneak peek at what we will cover:

● Structuring AI Data Schemas: Learn how to simplify your schema for AI by
explicitly including vital business fields and excluding confusing technical keys,
staging tables, or free-text columns.

● Synonyms & AI Instructions: Stop AI hallucinations by providing business-specific
phrasing and explicit instructions (e.g., teaching the model that "Client" means
"Policyholder" or guiding it on specific date logic).

● Verified Answers: Discover how to bind your report visuals to specific trigger
phrases (like "How many active policies do we have?") to ensure users get
consistent, validated, and organisation-approved answers every time.

● Data Privacy & MCP Servers: We're addressing the big elephant in the room. Learn
how Model Context Protocol (MCP) servers securely connect your semantic model to
external AI tools, and how to navigate the privacy implications of local, self-deployed,
and managed models.

Don't let your AI deployments fail due to poor data preparation or security blind spots.

Join us to learn how to build an airtight, AI-ready semantic layer!

Are you interested in the DAMA CDMP Certification... But not completely sure what it is all about, or how to get it? Whether you're just starting your data management journey or looking to advance your career, this session is for you.

We'll cover everything you need to know about the CDMP certification process, certification levels, exam tips for success, effective study plans, and career pathing assessments.

Don't miss out on this fantastic opportunity to boost your data management credentials and set yourself up for success.

Secure your spot now and take the first step towards becoming a CDMP-certified professional! 🎓

Email Debbie social@modelwaresystems.com for more information

AI governance frameworks are evolving rapidly — but a critical gap remains.

While standards such as ISO/IEC 42001, NIST AI RMF, and the EU AI Act provide important governance principles, they still lack operational mechanisms to continuously monitor and manage data quality within AI training corpora.

Join DAMA South Africa for an insightful webinar with Dale Rutherford, who will explore this often-overlooked governance challenge and introduce practical frameworks designed to detect, measure, and remediate data quality degradation, bias amplification, and model drift in AI training data.

This session will introduce the BME (Bias, Misinformation, and Error) Metric Suite and the MIDCOT methodology, providing a structured approach to monitoring corpus health and identifying risks before they propagate into deployed AI systems.

A key theme of the discussion is AI autophagy — the self-destructive cycle where models trained on AI-generated content reinforce bias, misinformation, and declining information quality.

This session is Part 1 of a two-part series, focusing on Monitoring and Detection in AI corpus governance.

📅 Date: 16 March
⏰ Time: 16:00 (GMT+2)

If you work in data governance, AI governance, data quality, or data management, this session will provide valuable insights into how data management practices must evolve to support responsible AI.

🔗 RSVP link will be in the comments

#AI #AIGovernance #DataGovernance #DataQuality #AITrust #DAMA #DataManagement #ResponsibleAI #AISafety

Medallion architecture has been a masterclass in vendor marketing. It has the scent of new lamps for old. A fresh name, polished diagrams, compelling colour coding.

And yet, layered data architectures themselves are not new. They have delivered real value for decades. Separating raw data from designed data from consumable data is not hype. It is a proven architecture.

The tension has never been the idea of layers, the tension has always been this:
- What data actually goes in each layer?
- What patterns are allowed in each layer?
- What governance expectations change as data moves upward?

Those arguments have played out in design reviews, architecture boards, data platform teams and governance councils for years. Data engineers debate transformations. Data stewards debate ownership and quality. Architects debate design.

After feeling this frustration for decades, I decided to try and make it slightly better.

I created the first iteration of an open source checklist. A structured, explicit way to describe the context of what each layer means in your organisation, what is allowed to happen where.

In this session, I will take you through the open source Data Architecture Layers Pattern Checklist. We will explore how it can reduce ambiguity, expose assumptions, and create a shared language between engineering and governance.

And then I will try to persuade you to help me iterate it further.