Pega’s annual user conference kicks off today in Las Vegas, with CEO Alan Trefler taking the stage for his first in-person keynote since 2019, sharing the company’s vision for how artificial intelligence will be built into the vendor’s Infinity platform. explained. Especially the role of generative AI. It’s worth bearing in mind that not all vendors take the same approach and it’s still in its early stages, but Pega’s AI pitch is firmly grounded in the company’s existing business logic architecture, offering a “bonus” ” can also be considered.
That’s an apt description, because many of Pega’s plans include its “situation layer cake” technology, a center-out architecture that unites the knowledge, processes, workflows, and case management that underpin an organization. Trefler describes it as a “Google index” for all the work we do as a company, and he created a “process fabric” that connects front-end development and back-end live data all in one place. Allows decision engines to be created.
The “garnish on the cake”, or layer cake in this example, will be artificial intelligence and automation. Pega’s goal is not to generate code using a Large Language Model (LLMS), as other software vendors in the market have preferred, but instead use LLMS to develop enterprise business logic. and strengthen the decision-making engine.
Trefler said of the layer cake architecture:
And this way of thinking led to the first concept of the layer cake. The idea of creating a structure to organize processes, rules, all elements, interfaces and technologies in a way that brings order. This allows you to share things, but also allows you to compartmentalize certain parts so that you can establish common rules and processes across your organization.
He argues that a decision-making “engine” built on this layer cake concept creates the ability to capture and act on intelligence, but rather than think of it as a monolith, it is a distributed network of connective tissue (process structures). ), he added. organization). This makes it possible to consider regional differences within the framework of a global organization. Pega argues that business should run as a process fabric.
This is a somewhat difficult concept to grasp (I think Pega wants to show that their architecture is uniquely positioned to take advantage of this new technology, but this doesn’t come down to soundbites). is difficult). The image below, taken from today’s keynote, gives a better idea of what this will look like in practice.
See how the Center’s situational layer cake, which connects to the front end via Pega’s DX API and integrates live data feeds from the back end, uses case management, workflow automation, and AI-powered decision-making. You can check if it is built on .
Commenting on how generative AI, and AI in general, is used in the world of Pega, Trefler reinforces the message that architecture matters. He said:
Frankly, if you’re using smaller apps and systems that aren’t very sophisticated, it’s probably fine. But the problem is that the logic is embedded in this code. More code, and inevitably more bugs. The reality is that more and more apps are not connected. Already, some people are finding themselves accumulating a lot of very bad code, which belongs to the technical realm and is called technical debt.
I use it to enhance layer cakes. We are not using it to generate code, we are using it to generate rules, business logic and AI principles into cakes and their structures. So there’s a place where you can see and contemplate what it’s doing. In fact, show the auditor what they are doing without a flood of prompts. Imagine what it would be like.
And this is the basis for the most important thing. That means you can change it, and you can weave it in and control it across your business. to permeate it throughout the enterprise. It’s a big idea, but it’s made possible by our own architecture.
real example
Chief Product Officer Kerim Akgonul took the stage to explain how some of the new generative AI capabilities are useful within the context of the Pega platform. Akgonul and another of his Pega colleagues showed how companies can quickly build a mortgage application workflow, or app, by simply typing “create workflow for mortgage application” into the system. I had a live demonstration of him.
A useful aspect of the demo was that while the workflow was set up quickly, people needed to understand what was required to make the workflow work. This includes case lifecycles, personas, data models, integrations, sample data, user interfaces, languages, and operational insights.
The difference is that instead of having a group of smart people sit in a room and build all this manually, they follow the initial prompt to say ‘Please add people who are participating in this workflow’ or ‘Give me an example of a house ” and simply follow up. Application workflow with sample data” or “Can I build my home her application in Turkish?”. Some images below show how this unfolds.
The demonstration was impressive and it was easy to see how it would save time. Building workflows, matching data, and language translation were all quick. Akgonul was honest that this wasn’t always 100% perfect for him, but that he could get close enough to be able to make the necessary edits and adaptations with human intervention.
And it was equally clear to see and understand what the AI was doing. No different components or layers were hidden from view. But the model seemed to understand what should go where.
Combine this with Pega’s newly announced process mining capabilities, and you’ll be able to ask AI models questions like, “How can we speed up this process?”, with some examples of where fixes can be applied. provided and get suggestions. Very interesting.
key infinity update
Today, Pega also announced various updates to its Infinity platform that unifies customer engagement, customer service and automation capabilities. It’s also worth noting. These include:
- Constellation UX – Pega’s design system, Pega Constellation, aims to deliver an engaging, accessible, and intuitive user experience in a low-code architecture. Constellation includes built-in design components, templates, and patterns.
-
Process AI Case Classification – AI capabilities predict classes and categories of specific cases, classify data to implement client-selected data classification categories, and route and assign work to appropriate individuals. Additionally, relevant context is provided to users to help them understand the situation and resolve cases faster.
-
strategy optimization – New conversion modeling uses adaptive models to help determine the propensity of customers to perform specific actions, such as purchasing or completing onboarding. In addition, new capabilities for testing and deploying machine learning models aim to help users determine success by conducting field experiments with specific models for high-value use cases and complex interactions. help to do.
-
Pega Process Fabric Connector – Pega integrates Pega Customer Service and Pega Process Fabric to make it easier for customer service agents to prioritize and manage work across the enterprise with connectors that unify work discoverability and prioritization Aiming to be Its purpose is to allow a customer service agent to view work and take action from any of her Pega applications.
my view
There’s a lot to digest on day one of Pega’s user event. Given the architectural focus of Pega’s AI opportunity pitch, I suspect that the opportunity itself is probably being somewhat lost on the technical side. Technical details are important, of course, but what really brought this story to life was the use case we got on stage today. At this point, buyers are probably wondering what the results of AI will be and how it can be used in practice. Of course, architecture is important to achieve this, but I think some of the theoretical examples of process fabrics and center-out decision engines can be difficult to understand when connecting them to AI components and technologies. increase. possible ones.
Maybe it’s just me, but I’m wondering if there’s a more concise way to understand this without making it feel so abstract. That said, the use cases highlighted on stage were compelling and I was genuinely impressed with the demonstration. Creating a mortgage application took a fraction of the time compared to the manual work required in previous versions of Pega. It made me realize how powerful the combination of prompting and human system understanding is. . I look forward to speaking with management later this week to delve deeper into all of this. For more information, keep an eye on diginomica.
