Wael Taha - Vice President of Enterprise Architecture at Brown Brothers Harriman

That is a very diverse journey.

I especially appreciate the chapters between the human-centric consulting side and the technical deep dive into the data and architecture world, which is such a unique balance that is not the luxury for many who end up in this position.

So thank you for sharing all of that.

That's really impressive.

And as I said, it's very well balanced, especially in the world that we're in today, which leads me To start thinking around, given the current role and the current type that you play in the analytics and the intelligent business BI space and AI space and governance space, the technical side of the world, it made me think, traditionally, data management or the data side has been very much siloed or limited to the structured side of the world.

In your earlier journey, as you were walking through, you talked about you named knowledge and information management being distributed.

And that's something, interestingly, we are seeing over the last three, four years, especially thanks to GenAI, where this role has expanded to the unstructured side for a majority of the world, right? Like content, documents, knowledge.

information, context, and semantics.

So I'm curious how you are seeing this affecting your role or your organizational roles within your organization, how these roles are maybe getting expanded as a result, and how your organization is navigating that change or that shift.

Yeah, so recently in the last maybe three years, I am, you know, most excited about how JNI and LMs are fundamentally changing what is possible with enterprise knowledge.

I guess we are moving from find this document to answer this question using everything we know which completely transforms how organization can you know leverage their collective knowledge the old paradigm in my opinion was traditional knowledge management that is built on you know organization and retrieval right look semantically helped right and that's what we were like working on the last five six years it helps retrieval we can map relationships create ontologies build knowledge graphs that would get you a lot of metadata, a lot of information about your data, but the census still required human effort, right? So a lot of human effort, a lot of knowledge, a lot of deeper knowledge, a lot of data stewards, people who understand their data.

You need to search multiple systems manually.

You need to read through documents yourself.

You need to synthesize insights in your own mind, on people's mind.

connect disparate pieces of information in spite that we have semantic layer, in spite that we have the ontology and all of this, what we can call like this kind of metadata layer or semantic layer.

I think within your paradigm and in the introduction of RAG, retrieval augmented generation, coming with the AI and the vectorization of information and stuff like that, systems don't just find relevant knowledge, they synthesize it.

Analysis can now ask questions in natural language.

So it retrieves information across systems, understands semantic relationships.

My personal experience is that an inspired semantic layer was great, but you need people with domain knowledge to be able to touch the data comfortably, especially with highly regulated environments.

Confidence And of course, we are on a cutting edge that we are evolving.

So I cannot say that we have managed, I would say, great success stories yet.

But we are really evolving fast into what I can say, being more confident about your data.

Because the ability of the RAG systems and in Gen AI to connect information and give you confidence.

Because now I, oh yeah, this data point is exactly what I need.

This is kind of insight that is based on these facts that I can then be more confident about it rather than just shooting in the dark and try to figure out things myself or come back to go and ask an expert, which are always busy and they are not accessible.