Most people might not even know that the arms race in creative AI has started.
GenAI can be used as an assistant to help people find their way around dashboards, messaging apps, and datasets. It can also summarize papers and chats and answer natural language questions.
It’s what everyone does.
A lot of people talked about Pega’s Knowledge Buddy this week at PegaWorld. This is a helper that is trained on an organization’s own data and also connected to an LLM, in this case OpenAI on Azure. Thoughts from Pega’s CTO Don Schuerman about their own answer were very clear: “I think Knowledge Buddy is well-designed and solves a lot of enterprise problems [but] I’m not under the impression that it’s going to be the only RAG-based product that’s sold.” “Everyone has one.”
Everyone has one. Mostly true. Adobe has an AI helper. There is an Einstein Copilot in Salesforce. HubSpot has a number of AI helpers, Microsoft has its own copilot, Oracle has a Digital Assistant, and SAP has a copilot named Joule. It’s a long list. It doesn’t take long to realize that having an AI helper is important, not because it makes you stand out, but because everyone else does.
Picture and words
Or, think about the genAI tools that have gotten a lot of attention: text and picture generators. A lot of people are interested in them because they’re easy for many people to get. This makes it possible for words and pictures to interact in a way that is truly unique. Anyone can become a maker, and the lines between writer and artist become less clear. No, I didn’t write that dull line in italics. I went to Google Gemini in a new tab and asked it to write the next line for the article.
I wonder how many of the more than 14,000 marketing technology companies don’t have some kind of genAI. Yes, we have it here. MarTechBot can answer questions about marketing technology because it was taught on our own archive. There is also a picture maker.
Text and picture generation is getting better quickly, but it’s important to remember that giving people the tools doesn’t have to be magical, no matter how magical the results are. It will seem strange not to be able to do things before you know it. Like a grocery store that doesn’t sell eggs.
But wait, there’s more!
The race to make AI smarter is over, right? These tools will get better over time, but they will still be available everywhere. They are the stakes in the ground.
I did not say “the AI arms race.” The arms race in creative AI is what I meant. It’s simple for people who don’t think about AI all day to think that creative AI is all there is to it. But that’s not even close to being true. In the real world, AI is used for many things that are not creative AI. Predictive analytics driven by AI, for instance, or classification AI like the kind that is used to make digital asset management systems tag files automatically.
In a broad sense, non-generative AI is scientific. It takes in a lot of data, processes it, and then makes something based on that data. It does not give you new information; rather, it tells you what the information would tell you if you were able to sit down and examine it yourself.
It is important to have this kind of AI. It can tell you about products or information. It can also tell a customer what to do next to help them get through a complicated trip. Do you know what that is? From what I’ve written about PegaWorld this week, you should know that “Pega is also, and perhaps primarily, a decisioning and workflow automation platform, driven by AI but not genAI.”
Making choices in real time
To understand Pega’s real-time decisioning, it helps to remember that the company began as a business process management service. It looked at problems with business processes and workflow as “cases” and used AI to suggest the best way to solve those problems. These were unique problems.
Somewhere along the way, Alan Trefler, founder and CEO of Pega, realized that this method could be used for CRM. Based on statistical analysis, AI could figure out which offer, message, or other form of engagement would work best for a customer at any point in their trip. These would be specific customers.
It looked like this was from a different world for a while than how Salesforce, Adobe, Oracle, and the other big players dealt with CRM. It looked like Pega was in a league of its own, with a name that not many people knew. However, it was successful at attracting big businesses, which were its target market, along with some governments and U.S. federal agencies.
As a matter of fact, Pega seemed to look down on the work of Salesforce et al., with their claimed batch processing and vague audience segments. Trefler told me in 2018 that there was “constant mediocrity.” He also said, “They’d love to buy us.” We’re making things very hard for them. We’re not for sale, though. We’re pretty publicly separate as a company.
What a strange thing it was to hear Salesforce talk about making decisions and the next best action in Einstein Personalization at this year’s Connections. It seemed like a recording of a Pega lecture from 2018. Does the fact that other sellers seem to be following suit make Trefler’s head explode?
He doesn’t really believe it, though. “It’s not easy to jump on board,” he told me. “It’s moving.” Even though he is glad to be validated, he still sees two “diametrically opposed approaches.” The way Pega does things is explained above. The other way is “Database marketing, SQL queries, lists of audiences, product-push…”Deep down, it’s there.
Making a bet
You can tell me I was wrong in two years when you read this again and again. In the past few months, I’ve heard over and over that AI is the future, but we don’t know what it will look like. I don’t think creative AI will rule that future. It will be as commonplace very quickly as email, online booking (remember going to the travel agent?) and Amazon book suggestions (those are powered by statistical AI, by the way).
If AI is going to really change the way businesses work, it will have to be used in some other way. Maybe improvements to AI that can predict the future, or maybe AI that is truly independent and can make business decisions without being watched and without putting the business at great risk. Still, we don’t know. What I think I know for sure is that text and picture generation will not seem as important once everyone does it (which won’t be long).