By Alfonso Quijano, Chief Technology Officer, Lean Solutions Group

Artificial intelligence has moved beyond experimentation. Across transportation, logistics and
nearly every adjacent industry, leaders have stopped asking if they should use AI and started
asking how to use it effectively. The trouble is, the “how” remains far more complex than the
hype suggests.
Recent research from MIT/NANDA, Gartner, and Bessemer Venture Partners highlights a
growing disconnect between the ambition for AI-enhanced productivity and its execution. MIT
NANDA’s study stirred debate by claiming that 95 percent of companies aren’t seeing
measurable ROI from AI investments. Gartner’s latest quadrant on AI maturity showed that
organizations operating in hybrid “human plus AI” models are outperforming those attempting to
automate too aggressively. And Bessemer’s 2025 State of the Cloud report found that while AI
now accounts for the largest share of venture investment, few companies have translated those
investments into clear, repeatable business outcomes.
Together, these findings tell a story: AI isn’t failing because the technology is flawed. It’s
struggling because most organizations are not yet designed to use it well.
The MIT figure – 95 percent of companies surveyed seeing no AI ROI – makes a compelling
headline. But it also oversimplifies the issue. Return on investment depends on how success is
defined. Is it faster turnaround times? Lower operational costs? Higher customer satisfaction?
Many organizations adopt AI with vague expectations and then conclude it “doesn’t work” when
the results don’t align with undefined goals.
In logistics, this ambiguity shows up in everyday scenarios. Companies implement AI-driven
automation for invoicing or document processing, expecting an immediate payoff. But when the
underlying process is broken, automation only accelerates the inefficiency. It’s the equivalent of
paving over potholes instead of fixing the road.
The more effective approach is to begin with diagnosis rather than deployment. Before
introducing AI into a workflow, businesses need to evaluate how the supported business
process functions, where data gaps exist, and what kind of outcomes matter most. Lean
Solutions Group clients sometimes ask, “Can you fix this?” A better question might be, “Should
this process exist in its current form at all?” Once that clarity exists, AI has a fighting chance to
deliver on its promise.

The Real Barrier: Organizational Design
Companies that succeed with AI decentralize implementation and empower teams to make it
their own. Those that fail often treat AI like a top-down initiative, controlled by a small leadership
circle or IT department, with little alignment to frontline workflows.

In practice, this means the technology is installed but not integrated. Employees don’t know how to use it, managers don’t know how to measure it, and leaders don’t know whether it’s helping or hurting. The solution is not more technology; it’s more intentional collaboration between people, process, and systems.
Implementing AI is less like flipping a switch and more like merging two companies. It changes
how teams communicate, how decisions are made, and what success looks like. Without
redesigning the organization around those realities ROI will remain elusive no matter how
advanced the tools become.
Gartner’s most recent analysis offers a useful framework for understanding how organizations
evolve along the AI maturity curve. On one end are those that rely almost entirely on human
decision-making, often burdened by manual processes. On the other are AI-heavy operations
that prioritize automation at all costs. But the highest-performing organizations are those that
operate in the middle, were humans and AI work in tandem, each reinforcing the other.
This hybrid approach aligns with what’s actually working across the logistics sector. In high-
volume but low-leverage tasks, like data entry, invoicing, or document auditing, AI can take the
lead, processing information faster and more accurately than humans ever could. But in high-
leverage environments, like customer service, operations strategy, or software development,
people still need to guide the technology.
AI can generate code or quotes in seconds, but it takes human expertise to define the right
outcome, check for accuracy and understand customer context. The most productive teams
today aren’t the ones that automate everything; they’re the ones that know what not to
automate.
Bessemer Venture Partners’ latest market analysis echoes this reality: productivity and speed
are not the same thing. Companies have historically measured efficiency by volume: how much
code, content, or communication a team can produce. But the rise of generative AI has blurred
that metric.
In many organizations, AI has made it difficult to tell who’s performing well and who’s simply
producing more. The result is what Harvard Business Review recently called “work slop”: a flood
of low-quality output that increases rework and confusion. More isn’t better; better is better.

Talent Is Not Being Replaced, It’s Being Redefined
The notion that AI will replace workers has proven to be an oversimplification. What’s actually
happening is a reconfiguration of roles. AI is absorbing routine, high-volume work, allowing
humans to focus on creative, analytical and relational tasks. This shift doesn’t eliminate the
need for talent — it changes what talent looks like.
The organizations seeing the most success are those investing in AI literacy across their
workforce. They’re training employees to use new tools and to understand their purpose.
They’re creating feedback loops where technology improves through human insight, and vice versa.

When AI becomes a collaborative partner instead of a distant command, adoption grows
naturally — and so does performance.
The market consensus is clear. AI’s potential is enormous, but realizing that potential depends
on how organizations design around it. MIT’s research shows the cost of ignoring structure.
Gartner’s frameworks illustrate the benefits of balance. And Bessemer’s analysis reminds us
that chasing output over value leads nowhere.
For logistics and transportation leaders, the takeaway is simple but urgent: success with AI
won’t come from adding more software. It will come from building smarter systems,
organizational, operational and human, that know how to use it.
Technology may power the next era of logistics, but people will determine how far it goes.