AWS re:Invent has increasingly become a signal for where enterprise technology is actually headed, not just what’s being promised. In this conversation from the show floor in Las Vegas, Pankaj Kumar, Global VP of Strategic Alliances and Head of the AWS Business Group at Kyndryl, shared a grounded perspective on what it really takes to move agentic AI from pilot projects into real production environments.

Rather than focusing on abstract AI potential, the discussion centered on execution, data gravity, and the operational realities faced by Global 2000 enterprises.

Inside Partnering is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Why So Many AI Pilots Stall

One of the core challenges Pankaj highlighted is something many partner leaders are seeing first-hand: an explosion of pilots with very little follow-through.

“There’s a lot of money being poured into agentic AI that start with the pilot, but a very small percentage of those pilot projects actually see ROI in terms of real world executable projects.”

The issue isn’t lack of tooling or ambition. It’s that most pilots don’t connect deeply enough to the systems that actually run the business. Without access to mission-critical environments and operational data, pilots remain isolated experiments.

The Role of Mission-Critical Systems

Kyndryl’s position in the market is unusual. The company manages complex, long-running systems for a majority of Fortune 100 enterprises, including a significant portion of the world’s managed mainframe environments.

“We manage more than two to three decades of mission-critical systems for our customers. I think we have access to data which very few in the industry have.”

That access becomes the foundation for making agentic AI useful. Instead of training agents on abstract datasets, Kyndryl works directly with the operational knowledge embedded in real systems, workflows, and infrastructure.

The Enhanced Agentic AI Framework

At re:Invent, Kyndryl highlighted its enhanced agentic AI framework, which combines several layers:

* Deep operational knowledge from managed services

* Proprietary platforms like Kyndryl Bridge with billions of data points

* Agent builders developed by Kyndryl

* Native integration with AWS services

“That repository of knowledge that we have is the critical fuel that you require to make an agent AI project successful.”

This approach is not positioned as a one-size-fits-all product. Instead, the framework is designed to be customized by region, industry, and customer maturity.

From Assessment to Execution

Another theme that surfaced repeatedly was execution discipline. Kyndryl works across the entire transformation lifecycle, from assessment and migration to modernization and operations.

“We are able to translate those pilot projects into real world executable projects.”

That translation is reflected in business outcomes. According to Pankaj, a meaningful portion of recent Kyndryl signings already include an agentic AI component, suggesting that customers are moving beyond experimentation.

AWS as a Strategic GTM Partner

The conversation also explored the depth of the AWS partnership. Kyndryl has aligned its go-to-market motion with how AWS now sells by industry rather than horizontal services.

“Our sales motion and our collaboration with AWS is in tune with how AWS is going to the market, what solutions they are taking to the market.”

This includes co-sell engagement across financial services, insurance, healthcare, automotive, and the public sector, where Kyndryl was a finalist for multiple AWS Partner of the Year awards.

Industry-Specific Execution Matters

One insight particularly relevant for partner leaders is the emphasis on regional and industry focus.

“Something that works in Japan could not be the same thing that we take to continental Europe or Australia.”

Rather than forcing global uniformity, Kyndryl and AWS customize plays based on local regulations, market maturity, and customer expectations, especially in regulated industries and sovereign cloud environments.

Consulting as a Growth Engine

Since spinning out as an independent company, Kyndryl has invested heavily in consulting capabilities to complement its managed services heritage.

“Our consult business is more than $3 billion, growing at about 35 to 40% year on year.”

That consulting layer allows Kyndryl to engage earlier in transformation conversations and stay involved through execution, which is essential for agentic AI initiatives that cut across infrastructure, data, and operations.

A Grounded View of What Comes Next

Rather than framing agentic AI as a sudden revolution, Pankaj described it as an evolution that depends on execution maturity.

“This would be something where Kyndryl fills a gap in the market.”

For partner leaders, the takeaway is clear: success in the agentic AI era will depend less on novelty and more on operational credibility, trusted data access, and tight collaboration with hyperscalers.

🎙️ Inside Partnering is a podcast for ecosystem builders, alliance leaders, and the people shaping the future of partnerships.

Let’s build the future of partnering – together.

📌 If you found this post helpful, would you please consider restacking it and sharing it with your audience? This spreads the word and keeps me interviewing and sharing content that will help you grow your partnership business and career.

Thanks for reading Inside Partnering! This post is public so feel free to share it.

🎧 Want more conversations like this?

💌 Subscribe to get new episodes and behind-the-scenes insights: insidepartnering.substack.com

Check out all 130+ episodes at InsidePartnering.com

🔗 Follow Chip on LinkedIn for daily partnership content and guest clips

Know someone Chip should interview? Send a quick email.

This is a public episode. If you’d like to discuss this with other subscribers or get access to bonus episodes, visit insidepartnering.substack.com/subscribe