Skills are the product: Inside Kyndryl's high-stakes approach to learning and development

Misha Santa Barbara spent most of her career at IBM, zigzagging across strategy consulting, design thinking, and business leadership rather than going deep in any single function. That breadth was by design.

"The opportunities were always there to learn something new," she says. "And I always jumped into a new area."

When IBM spun off its Global Technology Services division to form Kyndryl in 2021, Misha came with it. What happened next was a pivot she did not plan: the HR team asked her to lead learning and technical skills for the entire organization.

She took the offer.

What made her an unconventional hire for an HR function also made her the right one. In her own words, she had been one of HR's most demanding clients, frustrated by slow timelines, generic experiences, and a function she felt was too process-heavy and not human enough. Moving from critic to builder gave her a perspective most learning leaders never have.

Why speed and human experience matter most

Two things topped Misha's list of frustrations when she was on the business side: speed and authentic human experience.

"I just hated standing still and waiting for something to happen," she says. "Coming from business, I was used to 'good enough, let's go.'"

The second gap was subtler. She felt the human element was missing from HR programs, not in intent, but in execution. Too often, the focus landed on the process, the tool, and the approval chain rather than on how the experience actually felt for the person going through it.

She brought a design thinking lens to close that gap. Her goal: understand users, not just workflows. That philosophy now shapes how her team partners with technical leaders across Kyndryl's 70,000-person organization. More on building learning programs that drive real outcomes.

Skilling at speed: The business case at Kyndryl

At Kyndryl, the stakes around learning are unusually concrete. The company sells technology services, which means its people are its product. Skilled consultants generate revenue. Unskilled consultants cannot be staffed on client engagements.

"We are taking people away from their billable utilization, so it better be really good training," Misha explains. "There is a direct link to business results."

That accountability cuts through common L&D ambiguity. When Misha's team builds a learning program, the business question is immediate: did the people who went through it become more deployable? Did they support deals they previously could not?

This is the model more L&D leaders are being asked to adopt: connecting professional development skills directly to commercial outcomes rather than measuring completion rates and calling it success.

Piloting AI-powered learning: What worked and what they learned

Kyndryl ran one of the more ambitious early experiments in AI-driven learning: custom avatars built into sales training courses, designed to simulate realistic client conversations.

Rather than letting participants practice with classmates in breakout rooms, learners interacted with AI avatars fed detailed behavioral profiles based on real client scenarios. They had to extract the right information, ask the right questions, and earn a second meeting. If they did not get there, they tried again.

"People were like, this is really cool, can you give it to us so we can practice at home?" Misha recalls. The enthusiasm validated the concept. The engagement was real.

But the pilot also surfaced hard lessons. The first: experience quality is non-negotiable. If the AI interaction was clunkier than a practice run with a colleague, the extra investment was not worth it. The second: cost predictability was a bigger problem than expected.

"It feels a little like when the cloud was out there and everybody was moving to cloud and then you got the bills," she says. The token-based pricing model made it difficult to scale open access, so broader rollout was paused while the team figures out a sustainable model.

This kind of measured experimentation reflects how the best learning organizations operate. They prove a concept, measure it honestly, and build toward scale rather than chasing the technology for its own sake. It connects directly to how AI skills development works best when anchored to clear business problems.

Beyond role play: Agents for the everyday work of learning

Kyndryl's AI experiments extend beyond the sales training pilot. The team has built agents to help employees set development goals, navigate learning resources without getting lost in an overwhelming LMS, and prepare for coaching conversations.

These are smaller interventions with a different value: they reduce friction in the daily flow of work rather than replacing structured learning programs. Misha sees both as necessary.

"The potential is so much higher," she says of the agentic direction. "We're like in on it. We just need to figure out the personalization and how we can truly scale it at the right ROI."

That tension between personalization and scale is one of the defining challenges in modern talent development strategy. Misha's framing captures it precisely: 'personalized scale sounds like an oxymoron, but AI is the way to do it.'

The three kinds of learners (and what each needs)

When it comes to driving adoption of new learning programs, Misha maps her 70,000-person workforce into three groups.

The first group learns without being asked. They wake up every morning and absorb something new. For them, the job is pointing them in the right direction.

The second group is rational and skeptical. They need to understand the why before they engage, and they respond well to clear communication, leadership support, and practical change management.

The third group will not participate regardless of what is offered.

For a professional services organization, skills are not optional. They are the job. This framing is a useful model for any L&D leader trying to build a culture where learning is expected rather than encouraged. Explore how learning and development programs can be designed to serve all three groups.

What this means for L&D leaders

Misha's story is not just about technology. It is about what happens when someone who genuinely expected more from L&D gets the chance to rebuild it from the inside.

The mindset shift she pushed for on her team, moving from request-takers to consultants, from note-takers to strategic partners, is the same shift HR leaders across industries are being asked to make. And the AI tools she is piloting are not replacing that shift. They are amplifying it.

The blank page problem in L&D is real. When you know your people are your product, you cannot afford to stay there.

Listen to the full episode of Skilled for more on AI-powered learning, scaling personalized development, and building L&D strategy from the inside out.

Resources:

• Connect with Misha Santa Barbara on LinkedIn
• Learn more about Kyndryl
• Explore Growthspace for precision skill development at scale