The Art of Interview for AI Operations at data.world

In my role as VP of AI Operations at data.world, I've found that the difference between successful AI initiatives and those that fail often comes down to one critical factor: choosing the right starting points.

No matter how advanced the technology, AI solutions typically fail when applied in a vacuum—without first understanding the unique workflows, bottlenecks, and pain points that employees experience daily. That's where the AI Operations interview comes in.

What is an AI Ops Interview?

The AI Ops interview is a structured approach that allows AI teams to act as detectives and consultants, uncovering valuable AI opportunities by engaging directly with employees. It's not about pushing AI for AI's sake; rather, it's about listening, identifying inefficiencies, and finding the areas where AI can deliver real, measurable impact.

A well-executed interview provides actionable insights, reveals opportunities that might otherwise be overlooked, and helps align AI investments with actual business needs.

The 6-Step AI Ops Interview Process

At data.world, I follow a clear structure to build relationships and trust with stakeholders:

1. Building Rapport

• Begin by fostering a comfortable, open conversation
• Ask about the interviewee's role, department, and experiences
• Examples: "What led you to this role?" or "What do you enjoy most about your current position?"

2. Understanding AI Perceptions

• Ask an open-ended question: "What are your thoughts on AI?"
• This helps gauge their level of understanding and sentiment toward AI
• Allows me to identify which persona I'm speaking with (Energetic Emma, Skeptical Sam, Cautious Chris, etc.)

3. Defining the Department's Mission

• "What is the primary objective of your department?"
• "How do you measure success?"
• "What is the most common reason for failure?"
• "If you could wave a magic wand and change one thing about your work, what would it be?"

These questions help establish key success metrics and identify where AI can provide the most value.

4. Understanding Critical Processes

This is where the real discovery happens:

• Ask the interviewee to walk through their department's critical tasks step by step
• Utilize visual collaboration tools like Miro boards or whiteboards to map out workflows
• Look for inefficiencies, bottlenecks, and repetitive tasks where AI could optimize performance

For example, at data.world, we often find manual data entry across multiple systems leads to errors and inefficiencies, making it a prime candidate for AI-driven automation.

5. Exploring Potential Impact of AI Improvements

• "What would happen if we eliminated the most common points of failure?"
• "How would your team's productivity and job satisfaction improve if these challenges were resolved?"

This step helps frame AI solutions in the context of tangible benefits for the interviewee's team.

6. Closing Without Immediate Recommendations

• The interview should not end with AI Ops making on-the-spot recommendations
• Instead, take notes, systematically document key insights, and categorize findings
• Compile insights into a templated report delivered back to the interviewee
• After reflecting deeply, analyze the best AI (or non-AI) solutions to propose later

The Power of Visual Mapping

One technique I've found invaluable is visual process mapping. By creating a shared visual representation—whether on a whiteboard, Miro board, or even a simple sketch—both parties ensure they are aligned in their understanding of workflows, inefficiencies, and opportunities for AI intervention.

Why Visual Mapping Works:

1. Ensures Alignment on Process Details
   Employees often describe workflows verbally, which can lead to gaps in understanding. A live, mutually built process map ensures a shared, accurate representation.

2. Reveals Hidden Inefficiencies
   Mapping a process visually helps surface bottlenecks, redundancies, and gaps that might not be obvious in a verbal discussion.

3. Accelerates Post-Interview Analysis
   Instead of reconstructing an interview from notes alone, the AI Ops team can refer to the mapped process, saving hours in documentation.

4. Improves Stakeholder Buy-In
   When presenting findings to leadership, a clear process map helps illustrate why AI solutions are needed and how they fit into existing workflows.

Avoiding "AI for AI's Sake"

With the excitement surrounding AI, there's often a rush to apply it to every problem, regardless of whether it's the best solution. At data.world, we take an operations-first approach to avoid this trap:

• Process Improvement First: Sometimes a simpler process improvement can resolve the issue without AI
• Education and Training: Many "AI problems" are actually educational challenges
• Hiring or Role Adjustments: Sometimes the right answer is adding human capacity
• Verifying True Need for AI: Ask "Would this problem still exist if we improved our current processes?"

Real Business Impact, Not Hype

The best AI initiatives begin with a well-defined business problem, not with a technology looking for a use case. By maintaining a disciplined approach to evaluating AI use cases, we ensure our investments drive tangible value rather than chase trends.

At data.world, my team acts as the bridge between technological advancements and practical business applications, ensuring that AI adoption remains strategic and sustainable—delivering real value to our customers and our organization.

This post is adapted from my upcoming book on AI Operations, where I explore in greater detail how organizations can effectively integrate AI into their operations.