HEB also implemented Boomi DataHub (MDH) to deepen collaboration and precision. With MDH, HEB can comprehensively manage its digital assets through three abstracted layers: people, projects, and equipment.
“By establishing boundaries between the three data archetypes, ERP data accurately feeds through and lands where it should locally, and vice versa. For example, new hire information now pulls in our Boomi ‘people’ bucket and trickles up to the ERP when minimum data completeness is reached,” said Van Der Walt.
“We’re becoming a digital-centric business and not just a construction company that has data. This means we aren’t just installing shiny new tech; we’re giving systems meaning by connecting them around the stable enterprise data core. This is especially relevant to our AI plans, because you cannot have AI without IA (information architecture). With Boomi, we’re now ready.”
Future plans for AI
Although a little behind in the AI journey, HEB has overcome most of the hurdles already. “We now have 25 million records, and the data is clean,” said Van Der Walt. “So, it’s a happy coincidence that Boomi has released its AI offerings. There’s no need for us to undo anything, so we plan to use Boomi fundamentally for our AI use cases.”
For instance, HEB intend to use AI agents to predict with 95% accuracy in maintenance.
Currently in the stage of educating people within the organization about AI, he sees agentic AI operating in hybrid human+AI mode at least for the next two years.
Boomi CEO Lucas emphasized that human-centered automation is key in an era of AI-driven automation. That’s why, while applying AI to make things easier and better for humans, we need to put humans in control over automation.
Van Der Walt concurred: “In the same way we have HR systems to help manage people and evaluate performance, similarly we see Agent Control Tower doing that for AI agents, including governance. No matter how forward-leaning we are towards AI, we need human-in-the-loop to ensure we do AI right. We cannot rush into it, and must provide the necessary guardrails so humans will still be in control.”
He added: “Educated trust is proven trust. We as humans operate based on trust, and the current mistrust in AI requires education. Governance and monitoring are part of building this trust. Otherwise, five years later, we’ll be looking into how to undo AI. We need to slow down to go faster!”
Real change can only happen at the speed of trust.