Grygo is the chief content officer for FTF & FTF News.
I have often had moments when I am working on a story when a source dives into a great subject that I cannot incorporate into the story that I have planned.
Such was the case for the story I just filed about Gresham Technologies and its strategic alliance partner Cognizant Technology Solutions, and how they have developed an artificial intelligence (A.I.) and machine learning (ML)-based way to investigate the causes of repeated trade breaks for an operations team at a large firm. The story can be found here.
While I was interviewing the sources for the story, one of them — Neil Vernon, chief product and innovation officer at Gresham — mentioned the potential of large language models (LLM) technology.
First, we must define LLM, which Amazon Web Services says on its website “are very large deep learning models that are pre-trained on vast amounts of data. The underlying transformer is a set of neural networks that consist of an encoder and a decoder with self-attention capabilities. The encoder and decoder extract meanings from a sequence of text and understand the relationships between words and phrases in it.”
So, back to Vernon who says that Gresham is “finding that machine learning strategies where we can leverage a large language model [LLM] — they are working really well in our proof of concepts.”
Gresham clients “create very complex queries into our data because we’ve got a lot of data and people want to understand that data. We found that large language models can really help simplify the question that the user asks. And the large language model can guide the user and sort out ambiguities in the user’s question. So we’re excited by that and you will see in releases later this year with large language models embedded to make searching easier,” Vernon says.
In fact, LLM technology may be a good fit for helping clients sort out the Gresham documentation.
“Our documentation is somewhat dry and doesn’t really help you build reconciliation. It will tell you what a particular thing is, but if we’re honest, it’s not that helpful,” Vernon says.
“We can bring our documentation to life by adding a large language model into it and we think that will simplify some of the more complex parts of our configuration experience by bringing the documentation to life,” he says.
Beyond documentation, “in terms of building new reconciliations, it is easier to express the requirements through a large language model and have the large language model turn the requirements into configuration,” Vernon says. “And that’s an easier thing to do than to use our current studio approach.”
Neil Vernon
“So, we are doing proof of concepts here with large language models, but we think that it will be more of a conversation to build a reconciliation than using a traditional kind of studio. And we actually think that’s going to be a mind shift,” Version says.
“If I look at our millennials that are using the products, they are much more used to having these longer chat-based kind of approaches … And we think this chat-based approach could be really interesting for configuration and maintaining configuration,” he adds.
My guess is that if Gresham is already getting encouraging results from LLM then other providers may also be investigating LLM.
To say the least, we will definitely be watching the LLM, data, and operations space.
Need a Reprint?