Three years of enterprise AI investment, and most companies are still waiting for the payoff. Boomi LP thinks it has found it, unveiling Boomi Companion, a collection of open-source agent skills that lets developers build, deploy and test fully configured Boomi solutions through natural language instructions.

The announcement reflects a broader shift now taking shape across the industry, with the move from AI hype to hands-on value arriving not through better chatbots or smarter dashboards, but through tools that let developers build production-ready enterprise software with natural language instructions. As agentic AI moves from experimentation toward operational scale, the companies that have invested in platform-native AI capabilities are finding themselves well ahead of those that waited, according to Matt McLarty (pictured), chief technology officer at Boomi.

“I think that since ChatGPT exploded … everyone’s been seeing the inevitability of the AI revolution,” McLarty said. “But in practice, a lot of enterprises are [saying], ‘It’s hype, but I’m not seeing the value yet.’ The agentic engineering space  is really more refined than anywhere else yet. That’s what we’re seeing with our customers and partners — they’re [saying], ‘We were excited, we’ve been excited for three years, but we’ve also been wondering when’s the payoff?’ This is the payoff.”

McLarty spoke with theCUBE’s John Furrier and Gemma Allen at Boomi World 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed Boomi Companion, the future of agentic engineering and how enterprises should balance deterministic and probabilistic processing as AI reshapes the software development lifecycle. (* Disclosure below.)

Boomi Companion and the rise of agentic engineering

Enterprise developers have long had to choose between the speed of low-code tooling and the power of full-platform configuration — a tradeoff that Boomi Companion is designed to eliminate. Launched at Boomi World 2026, Boomi Companion is a collection of open-source agent skills built on Anthropic PBC’s open standard that allows any AI coding agent — including Claude Code, OpenAI Codex or Microsoft Copilot — to build, deploy, test and diagnose fully configured Boomi solutions through natural language, McLarty explained.

“As opposed to a typical old-fashioned code generation thing that might come up with a little skeleton thing, ‘Hey, here’s what you might do,’ and then you have to go fill in the details, Companion actually builds a fully robust solution,” McLarty said. “It deploys it, it tests it, it checks its results, it iterates through, it learns, it talks to the user. It’s one of those things that it’s almost a little too magical in terms of what it can do.”

The key distinction underpinning that “magic” is between agentic engineering and what the industry often calls vibe coding — a term describing hobbyist-style, shot-in-the-dark generation that produces brittle output, according to McLarty. Companion, by contrast, was built by an expert Boomi builder as a digital twin of deep platform expertise, embedding best practices directly into the agent’s skills so that every solution it generates is structured, scalable and production-ready.

“It’s not just barfing out a bunch of bespoke code and saying, ‘Here’s a big blob, go implement this,'” he said. “It’s actually built on best practices of a tried-and-true enterprise platform. The graphical interface takes on a new role — if I do build all this stuff, what did I just build? We have a way of showing how it’s built, structuring it, and you can still tweak it that way if you want. It’s like a bidirectional best-of-both-worlds.”

Beyond the immediate build capability, McLarty sees Companion as the foundation for a broader architectural rethink that every enterprise software team now faces. Drawing on his IT Revolution article applying Daniel Kahneman’s “Thinking Fast and Slow” framework to agentic systems,  the core challenge is not replacing deterministic processing, but knowing when to augment it with probabilistic reasoning — and designing systems that do both deliberately, McLarty explained.

“Getting the balance right between deterministic processing and probabilistic processing is the biggest architectural challenge that’s going to hit the software architecture,” McLarty said. “I think it’s really a new paradigm of software architecture. We went through object-oriented, we went through service-oriented, we went through API-first, microservices — this is a bit of a bigger leap, but it’s the prevailing software architecture practices that introduce probabilistic processing into real-time workloads.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Boomi World 2026:

(* Disclosure: TheCUBE is a paid media partner for Boomi World. Neither Boomi, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE