Integration.app Uses LLMs to Connect Apps
While system integration as a service isn’t new, the use of large language models could be
What was once considered a high value business model of systems integration in the software solutions business, might just become a simple task that can be accomplished with the flick of a few switches – or should we say click of a few buttons on the computer keypad! A new startup has raised $3.5 million by connecting a long-felt need with what could be an AI-led solution.
That is, if one can broaden one’s vision to term large language models (LLMs) as the core of AI innovations that are claimed as such. The company is called Integration.app and has managed to unwrap its solution with a very small team led by Daniil Bratchenko, one of the earliest employees of DataRobot, one of the earliest to identify AI as a game changer.
That Bratchenko’s experiments have resulted in his startup actually managing to hit the ground running to own the white space of SaaS-based apps and services not really talking to each other is intriguing. However, what reduces this intrigue is the use of LLMs to automate some aspects of the software integration process itself.
An AI-based systems integration platform
A report published by TechCrunch quotes Bratchenko as suggesting that his company helps engineering teams to integrate their products with every other SaaS that customers use. So, he prefers to call it the universal integration platform with application-specific customization that reduces the time required to regularly maintain tens or hundreds of such instances.
As an explainer, Bratchenko uses ChatGPT-like models in its connector builder tool that uses large language models to understand API documentation in order to help the engineers figure out easier ways to integrate specific apps. This ensures that customers seeking to integrate apps and services can actually build many connectors within a short period of time.
As an example, he notes that his company can help port over customer data from a CRM platform quite easily as the Integration.app will automatically map the APIs and UIs from across multiple apps and spew out several connectors at one go. This would save large amounts of time and effort of the developer community.
Cutting down time and effort is the key
The idea of working on such a product in stealth mode came up based on Bratchenko’s understanding that IT departments of every enterprise considers integrations while acquiring new software or another entity. With each division requesting multiple integrations across their customer / user base, the engineering teams are often hard pressed to create them.
So, while there are tools and more to build integrations between SaaS apps and services, getting them to seamlessly talk to each other takes time and effort. Especially since Statista data confirms that on an average companies use more than 130 apps in their business – a task that could quickly spiral out of control if not heeded initially.
Mind you, there are others in the field already
Of course, Bratchenko’s startup isn’t the only one in the iPaaS (integration platform as a service) line of business, which is currently estimated to be worth more than $3.7 billion. The likes of Boomi, MuleSoft (owned by Salesforce) and Digibee are already ensconced in this space, given that several large enterprises have already invested in the iPaaS space.
That Bratchenko used the money from his sale of movie recommendation engine Kionobaza.tv to Yandex to start-off his experiments suggests in spite of his knowledge of AI and LLMs, the story had to be kept under wraps. Why? Because others like Zapier have introduced AI-led tools that allow users to map out app-to-app workflows.
However, the $3.5 million funding from Crew Capital, Seedcamp and Cortical Ventures is proving that the idea was indeed something to invest in. Bratchenko says the money would be used to bolster go-to-market efforts and add to his 12-member team size. Integration went commercial four months ago and now has “dozens of paying customers,” he says.
And are there any challenges, other than growing competition? He believes Integration needs to engineer a shift in mindset among IT teams to leverage embedded integration technology though he believes his team can swing it – sooner than later. And when they (and others of their ilk) do so, we can finally say RIP to the much-vaunted System Integrations business.
