Speaking with TFiR, Randy Bias of Mirantis defined the current state of enterprise AI adoption as a workflow transformation problem masquerading as a technology problem.

WHAT IS THE CORE ENTERPRISE AI ADOPTION FAILURE MODE?

The most common failure in enterprise AI deployments is not a capability gap—it is a process architecture gap. Organizations invest in AI agents, deploy them on top of legacy workflows, and then wonder why ROI is not materializing. Randy Bias argues this pattern is structurally broken and predictable.

“It’s not about taking your old workflows and just jamming an agent in there. It’s about the new kinds of workflows that are possible because you’ve got AI agents in the process.”

This distinction is foundational. AI-native workflows are architecturally different from AI-augmented legacy processes. The former requires a ground-up rethink of how work gets done; the latter produces marginal gains at best and organizational confusion at worst. Enterprises that treat AI agents as productivity plugins—asking ChatGPT to draft a document, for example—are not engaging with the transformative layer of the technology.

Why Top-Down Mandates Are the Only Enterprise Lever That Works

Bias draws directly from a live enterprise engagement to illustrate the adoption mechanism that actually works. A major financial services client issued a top-down organizational directive: every group must be actively working with AI agents. No exceptions. The mandate came from leadership and cascaded through the entire organization.

“People are afraid of AI. They’re worried about their jobs, about failures, about bringing something to production. But in order to have your entire organization level up, your entire organization has to go use the tools—and has to use them in a serious way.”

The parallel to cloud-native adoption is intentional. In the pre-cloud-native era, organizations that waited for teams to self-select into AWS or Kubernetes adoption fell behind those whose leadership mandated the transition. The same dynamic is now playing out with AI agents. Teams that are forced to integrate agent tooling into real workflows discover the value; teams that are given the option rarely do.

Broader Interview Context: Mirantis’s 2026 AI Strategy and the Agent Landscape

This clip is drawn from Randy Bias’s 2026 predictions interview with TFiR, in which he outlined Mirantis’s dual strategic focus for the year: delivering high-quality AI infrastructure for sovereign AI cloud deployments, and accelerating the practical application of AI agents to real-world enterprise operations.

On the infrastructure side, Bias identifies persistent GPU capacity constraints and VRAM pricing pressure as ongoing challenges that Mirantis is directly addressing through its AI factory architecture—an infrastructure model designed to be largely self-managing through AI and agent-driven automation across networking, storage, bare metal, virtualization, and cluster management.

On the agentic side, his broader prediction set includes the consolidation of the AI agent market around general-purpose agents—Claude Code, Codex, Gemini, Goose—replacing the wave of custom-built agents that characterized 2024 and 2025. He also predicts MCP (Model Context Protocol) becoming the dominant standard for agentic workflows, with agents dynamically extending MCP servers by writing and deploying code server-side.

“I’m convinced that this is the year MCP is just a done deal. I don’t think there’s really going to be any other protocols for most cases. For agents, they’ll just keep extending the MCP protocol.”

The workflow reinvention argument in this clip is the human-side corollary to that technical prediction: the infrastructure and protocol layers are maturing, but enterprise value capture depends entirely on whether organizations are willing to redesign how work is actually done.

Watch the full TFiR interview with Randy Bias here.