What you need to know

• Integration remains one of martech most painful bug bears, say analysts and CMOs. But large language models fed with hundreds of millions of integration patterns offer a power accelerant.
• Boomi has been issued with a patent by the US Patents office which it (and other industry commentators) say will address some of the power common integration problems – getting apps from different vendors to more easily connect out of the box.
• According to Boomi CEO, Steve Lucas, the former CEO of Marketo, martech vendors aren't really trying to solve integration: "They're not giving you vitamins, you're just getting aspirin."
• Boomi's patent puts it in a solid position, and with Boomi GPT based on 300 million integration patterns, it is well placed. Yet rivals such as Workato, SnapLogic, and Tray.ai have access to their own vast database of patterns, so expect increasing competition. Ultimately, that's good news for all CMOs.

Is one of the most intractable problems in martech – integration – about to be solved? Or at the very least, broken open in a way that drives a great leap forward?

Mi3 Australia has learned Boomi, the global low code / no code digital transformation platform, has been issued with a patent that uses large language models to drive integrations between apps out of the box. Industry contacts Mi-3 spoke with last week in the US say it’s likely Boomi’s rivals, Workato, SnapLogic, and Tray.ai are also pursuing similar developments.

John Deeb, Workato ANZ country manager outlined almost half a dozen LLM initiatives to Mi-3 including its ReceipeIQ, which he says is trained on "a unique corpus of billions of data points from 700K+ community recipes that is used to recommend mappings, logic, and next steps." 

He also described Enterprise Automation LLM which he called the only secure, private, enterprise LLM that combines Generative AI and Workato’s own AI/ML models (RecipeIQ). "This powers features like Copilot, OpenAI connector, and AI Connect."

Brad Drysdale, Field CTO APAC, SnapLogic told Mi3 his company released SnapGPT in March 2023. He called this the first-ever generative AI solution for enterprise applications and the modern data stack. He said it leverages AI to quickly integrate and automate business processes using natural language prompts. Just this month the company also released GenAI Builder, which it calls the world’s first no-code generative AI application development product for enterprise applications and services.

"Product and IT teams now have the ability to rapidly add high-performing conversational interfaces to their most important digital experiences, changing them from basic tools into intelligent collaborative systems, creating an entirely new realm of capabilities for employees, customers, and partners, delivering unprecedented business value," he told Mi-3, in a written response..

“Instead of relying on multiple experienced python coders working for days and weeks to create a LLM-based solution, any employee – including those in finance, marketing, HR, and other business departments – can now use GenAI Builder to create solutions in a matter of hours. By using natural language prompts to create powerful generative AI solutions while simultaneously connecting disparate data sources, global businesses can save millions and accelerate their LLM projects tenfold.”

All of which suggests martech's integration challenge is increasingly likely to be solved outside of the martech ecosystem. For CMOs that can't come fast enough.

Martech integration, as we have reported in the past, is a rat's nest of cost and complexity, where the solution to every challenge invariably ends up being more martech.

And CMOs are getting sick of it. Just last week, we reported on Forrester’s latest global martech study which revealed integration has overtaken functionality as a key solution selling point. The researchers found while marketers can choose from countless martech solutions, successful implementation requires proper integration with existing martech stacks.

Yet 22 per cent of global B2C marketing decision-makers indicate process inconsistency due to poor integration of technology is one of the biggest challenges they face to achieving their marketing goals.That led Forrester to predict more than half of CMOs will prioritise martech ecosystem integration over marketing cloud dreams this year. And they will be more focused on functional consolidation than vendor consolidation as they strive to bring all of their operating tech capabilities in one place.

Boomi’s innovation, which is expected to be announced early this month, won't solve all of the problems – integration is a multi-headed hydra after all. But according to its CEO, Steve Lucas, the vendor offers a great leap forward by feeding its vast database of integration history into an LLM.

Lucas will be familiar to many marketers as the former CEO of Marketo, acquired by Adobe in 2018 for $4.75 billion. We spoke shortly after the Boomi was issued a patent from the US patent office.

“Like many companies, we are leveraging large language models and generative AI," Lucas told Mi-3. "But something unique Boomi did about 10 years ago is started collecting integration patterns. What this means is they're anonymised - we don't use actual customer data. But if someone builds a lead-to-cash process or an opportunity-to-close model, all of those integration patterns are stored in our cloud metadata database. 

"We've trained a large language model on them. So today, you can open up Boomi GPT and thanks to over 300 million integration patterns, type in a request to connect Marketo to Salesforce and multiple Salesforce systems, and go.”

Lucas says across his career, every single marketer he ever met identified system integration as a top three priority. "It's always a top-three challenge. It's the same for a seller, I'm talking about the CMO and the CRO," he says. In the past, the view was always to let the IT department or vendors figure it out.

"Here's what I know. Adobe is not going to figure out integration. Adobe is a sales and marketing technology company. That's what they are. And I know I worked there for a long time and at Marketo," Lucas continues. "The way these companies think about it is how do I connect my one system, for example Marketo, or my CDP to the bare number of systems I need to sell this product. They're not trying to solve the problem. They're not giving you vitamins, you're just getting aspirin."

By far, the most common use cases for Boomi today is connecting Salesforce and Marketo and Shopify. "I'd then add NetSuite to that. Why? Because people know they want to move from lead to opportunity to cash, whether it's B2C or B2B," Lucas says.

"The more time you can take out of that process, the more cost you can minimise, and the more precise and profitable the whole thing becomes."

Boomi's goal is to eliminate any consideration whatsoever for the system to process the data, the location - and more importantly back to CDP – to customer definitions.This is because customers are defined differently in different systems, Lucas says, flagging a major integration hurdle.

"To Boomi, it doesn't matter, we can literally connect to all of those systems and deliver a uniform way to look at that data. And we can do it in real-time," he claims.

Leap forward

Boomi isn’t solving all the problems of martech integration, which are often more than purely technical as we discussed in our review of CDPs. And of course, applications extend far beyond martech. But it’s a big move forward. Lucas, as the former Marketo CEO, is almost uniquely placed to describe it in terms CMOs understand.

Mi3 spoke to industry leaders with connections to companies including Boomi and rival Workato to gauge the significance of Boomi's IP and claims. These individuals were not authorised to comment and spoke anonymously.

One said: "The integration space is so massively large, even doing things that primarily improve out-of-the-box integrations where there hasn't been customisation - I can see how the company's [Boomi's] claim of not only doing that, but probably doing it faster with a generative AI layer, seems reasonable."

But there are caveats, starting with the idea the patent may be challenged. After all, Boomi’s competitors Workato, SnapLogic, and Tray.ai all have the capacity to create a massive history out of all the patterns generated by their customers, then feed them into a generative AI.

Next up, while out-of-the-box integration itself is a big enough problem worth solving, it gets more challenging as you have more bespoke data and integrations. The power of Salesforce and Adobe for many brands, for instance, is the unique capabilities customers can build into bespoke integrations. That’s also the edge of the sword that drives cost and complexity.

We don’t yet know the extent to which Boomi’s new innovation could accelerate that more complex body of work. While the company appears to address the issue of data integration between apps, complete martech integration is a much more complex beast. Just take the fact brands need to orchestrate data across multiple apps engaging with complex workflows and automation.

As one exec put it, "You don’t want your people switching between 50 different web browser tabs. Better that they are on one primary app and other things are surfacing up in their UI, which again is a very different kind of integration."

On top of that, there’s governance integration to ensures governance rules flow as seamlessly around the machine as the data itself.

Early days

We are still brewing the coffee pot on the LLM age. Boomi’s patent announcement, when it comes, could signal the emergence of much deeper integrations beyond basic out-of-the-box considerations. Those we spoke with are confident LLMs will play a greater role over time in addressing even the more intractable problems of martech integration.

When we get to a place where brands can use natural language to describe even the more bespoke aspects of martech integration, that will offer another huge accelerant.

Let's see.