Multi-agent orchestration is the destination, but for most enterprises, the road is blocked long before the first agent gets deployed by the quality of the data feeding those systems.

As organizations accelerate investment in agentic workflows, a persistent gap between AI ambition and operational reality is widening. Legacy systems were built for a world where humans supplied the interpretive logic that data alone could not. In the agentic era, that undocumented knowledge must be surfaced, codified and connected before multi-agent orchestration can deliver real return on investment, according to Patricia B. Moore (pictured), AI field chief technology officer and innovation lead at Boomi LP.

“It starts with trust. You have to have context in order to have that trust to have agentic solutions that are actually going to deliver value; you need your data to be connected, you need your systems to be automated,” she said. “The idea behind artificial intelligence is that there is intelligence, and in order to have intelligence, you need to have the right information.”

Moore spoke with theCUBE’s Gemma Allen at Boomi World 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how enterprises can clear the data trust hurdle and build toward multi-agent orchestration without abandoning the investments they have already made. (* Disclosure below.)

Multi-agent orchestration demands context before scale

The difficulty of moving from single-agent pilots to multi-agent orchestration at enterprise scale is reshaping how integration companies position themselves. Boomi launched Meta Hub earlier this year specifically to address what it calls the “tribal knowledge tax” — critical business definitions trapped in undocumented silos that cause agents operating without a human in the loop to make context-free decisions.

To illustrate the problem, Moore presented an example: Ask three departments to define an “active customer,” and the finance, operations and sales teams each return a different answer. Without a semantic layer to reconcile those definitions, agents will act on conflicting data, she added.

“When you have agents making decisions without a human in the loop, they need to understand the things that aren’t normally documented,” she said. “In this new agentic world, we have to think about, ‘How do we take that knowledge that has historically sat within our workforce and document it somewhere, whether it’s in business glossaries or elsewhere, to be able to feed that to the agents?'”

Boomi’s positioning in the market — with more than 30,000 customers and lessons drawn from years of API management — gives it a structural advantage as enterprises try to carry forward existing integrations into the agentic era, Moore added. The company is now applying the same governance principles it developed for application programming interface management directly to agent management, underpinned by a crawl-walk-run path to success.

“You need to pick those use cases that are going to be those quick wins to build confidence,” Moore said. “First, you’re going to automate — you’re going to learn from those automations so that you can agentify. Then once you build confidence in those agents, then we can start talking about multi-agent orchestration. You have to crawl before you walk, walk before you run, and we’re helping our customers at every stage of that journey.”

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. Sponsors of theCUBE’s event coverage do not have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE