As artificial intelligence adoption accelerates across supply chain and logistics operations, many companies are discovering that automation alone is not delivering the expected business impact. According to Alfonso Quijano, the problem is not the technology itself—but how organizations deploy it.

In this AI Reporter America Q&A, Quijano explains why operational context, workflow integration, and expert human oversight are becoming essential to successful AI implementation. He also shares how Lean Solutions Group is scaling its “expert-in-the-loop” model, where AI and human operators work together to improve accuracy, speed, customer experience, and long-term business performance across logistics operations.

What are the most common failure points or limitations you are seeing with standalone AI deployments in logistics today?

Most AI deployments fail because they’re deployed without proper operational context.

Companies expect AI to “just work,” but in reality, 95% of projects are delivering no measurable impact on profits (MIT Media Lab, “The GenAI Divide: State of AI in Business 2025,” July 2025) because they’re not tied to a specific workflow or business outcomes.

In logistics, that shows up as:

• Poor handling of exceptions
• Bad outputs from bad data
• Over-automation in customer-facing moments

The bottom line: AI without expert human oversight doesn’t scale—it breaks.

Can you provide 1–2 specific use cases or client examples where the Human + AI model has demonstrably improved outcomes?

The strongest results come from augmenting—not replacing—human operators.

Two common examples:

• Customer operations: AI surfaces data and recommends actions in real time; humans manage the interaction.
• Back-office workflows: AI executes repetitive tasks; expert humans handle exceptions and decision-making.

The pattern is consistent: AI improves speed and throughput. Human experts protect accuracy and experience.

How does LSG operationalize the “expert-in-the-loop” model at scale—what does the human intervention layer look like in practice?

This isn’t “human review after the fact.” It’s expert humans, augmented with AI tools, built directly into business operations workflows. AI handles high-volume tasks in real time, while human experts:

• Validate outputs
• Resolve or escalate edge cases
• Own the customer experience

Platforms like Lean Solutions Group’s LeanTek software platform embed this interaction directly into operations. Humans aren’t a fallback; they’re part of the workflow design.

What measurable KPIs or benchmarks can be attributed to this hybrid approach versus traditional automation models?

Hybrid models outperform across the metrics that matter:

• Higher accuracy
• Faster cycle times
• Lower cost per transaction
• Better customer satisfaction

That’s why a small percentage of companies are seeing outsized returns—while most are seeing none. AI alone cuts cost. Expert human + AI drives performance.

How do you see the Human + AI model evolving over the next 3-5 years?

AI will take on more execution, delivering more complex process automation with higher rates of speed. Humans will move further into oversight and orchestration.

But the fundamental model itself won’t change – and it becomes more critical as systems grow more autonomous. We’re moving toward hybrid workforces where digital and human operators work side by side. The companies that design that interaction well will win. Those that simply replace human operators with AI software products will continue to fall behind.

Are there any regulatory, ethical, or workforce implications companies should be mindful of?

Yes—and they’re already showing up. We’re seeing rising AI-related incidents, declining customer trust and increased scrutiny around data and decision making. The risk isn’t AI itself—it’s ungoverned AI. The fix is to structurally keep human experts in the loop where judgment, accountability, trust, and customer satisfaction matter.

Any final thoughts?

I always like to mention that the cost of technology delivery and innovation is going way down. That brings up the true bottlenecks in the process of actually getting such technology as AI implemented, which are change management, adoption, and other administrative tasks. As the cost barrier to entry for new technology gets reduced more and more, there are going to be more incumbents that are attempting to create new technologies. However, there are challenges that sit well outside the fact one has the ability to just be able to create technology. Among those are:

• How people actually interact with technology
• How to distribute it to the places where it's really going to drive impact without creating any type of shift or problems in the process

AI is no different.