DataBraid CTO Atif Khan Wins KDD Cup 2026 | Human-AI Innovation

The central idea behind PiTrace is both simple and profound.

Rather than building a system that produces answers from a black box, the team focused on building a platform where humans and AI agents work together to solve complex problems. As Atif explained during his presentation:

"We decided to write the analyst, not the analysis." 

PiTrace was designed as a human-agent workbench where AI performs the heavy lifting of discovery, analysis, planning, validation, and problem solving, while humans remain actively involved in judgment, interpretation, and decision making. The result is a transparent, evidence-based approach to AI that makes reasoning inspectable, explainable, and trustworthy.

As artificial intelligence continues to evolve, organizations face an important question:

How do you create AI systems that people can trust?

The Null Hypothesis team approached this challenge by building systems that expose evidence, validate conclusions, surface uncertainty, and invite human collaboration when judgment is required. Their work demonstrated that the future of AI is not simply about automation. It is about combining the strengths of intelligent systems with the expertise of human professionals. 

In a recent reflection on the competition, Atif highlighted four practices that proved critical to the team's success:

• Clear specifications that defined intent, constraints, and measures of correctness
• Using multiple AI models in complementary roles
• Treating implementations as replaceable while preserving evaluation frameworks and reusable skills
• Focusing on when human judgment should guide, validate, or correct an AI system

While this achievement was earned by Atif and his KDD Cup team, the ideas validated through this work strongly align with the vision we are building at DataBraid.

Insurance brokers operate in one of the most information-intensive industries in the world. Every day they must navigate complex relationships between clients, carriers, policies, claims, communications, and regulatory requirements. The future belongs to organizations that can transform that information into actionable intelligence while keeping humans firmly in control of the decision-making process.

At DataBraid, we believe the most powerful systems are not those that replace professionals, but those that help professionals perform at their best. The same philosophy that helped earn recognition on one of the world's premier AI stages is reflected in our broader mission to build intelligent systems that augment expertise, improve decision making, and unlock new levels of productivity.