Brad Tober

As AI has become increasingly central to marketing technology, I identified a critical gap in the approach being taken at Epsilon: while there was significant momentum around AI implementation, there was no cohesive philosophy guiding how these technologies should be integrated into user experiences. In response, I developed a comprehensive meta-design framework that positioned AI not as a replacement for human expertise, but as an amplifier of it. This framework sought to move beyond prompt engineering (in the context of chat-based user interfaces) toward more intuitive, integrated, and dynamic interfaces that would allow marketers to leverage AI while maintaining strategic control and brand alignment.

Business challenge

Epsilon was comprehensively exploring the integration of AI capabilities across its product suite. However, this effort faced several interconnected challenges:

• Fragmented implementation: AI features were being proposed in silos across different products without a unified vision for user experience.
• Expertise vs. automation tension: There was uncertainty about how to balance AI automation with the need for human expertise and judgment when making marketing decisions.
• Interface evolution stagnation: The industry (Epsilon included) was fixating on chat- and prompt-based interaction paradigms, without considering how interfaces might evolve beyond text-based commands.
• Risk of technical-first approach: Heavy investment in technically powerful systems risked underperformance if not matched with thoughtful integration into human workflows and decision processes.

These challenges threatened to create disjointed experiences that would fail to build trust, usability, or alignment with marketing professionals' actual needs.

Strategic approach

I approached this challenge by developing a multifaceted framework that addressed both philosophical and practical dimensions of AI integration.

White paper development

I authored "From Prompts to Principles: Meta-Design and the Evolution of AI Interfaces," exploring the concept of the meta-designer and establishing its relevance to AI efforts at Epsilon—positioning the designer as someone who shapes the systems and frameworks through which AI delivers outcomes.

View the white paper as a PDF (with username protected and the password you used to access this page)

Six-lens framework creation

I adapted Edward de Bono's Six Thinking Hats technique to create "Expertise, Agency, and Interface," a practical framework for evaluating AI-driven user experiences across dimensions including knowledge infrastructure, human perception, risk management, expertise scaling, interface evolution, and systems orchestration.

View the framework (with additional context) as a PDF (with username protected and the password you used to access this page)

Agentic AI product design POV

I collaborated with colleagues in senior leadership positions to develop a specific point of view on implementing agentic AI within Epsilon PeopleCloud, distinguishing between passive and agentic AI applications and defining clear use cases, human checkpoints, and performance metrics.

Throughout this process, my approach emphasized positioning AI as a tool for those who already possess expertise to work at a new scale or level of abstraction, rather than as a shortcut that bypasses the need for foundational knowledge.

Team and collaboration

This initiative required thoughtful cross-functional engagement across Epsilon's organization. I directly engaged with leaders of Epsilon's AI Center of Excellence, positioning the design framework as essential to, rather than an optional part of, the success of company-wide AI initiatives. This executive outreach was complemented by extensive collaboration with product managers across different areas to understand their specific AI implementation challenges and ensure the framework addressed real-world needs.

To ground this work in practical realities, I drew on insights from my extensive work with internal users, external clients, and analyst relations conversations conducted over three years. These interactions provided invaluable perspective on how AI capabilities would need to mesh with existing workflows and expectations. Throughout the process, I positioned myself as an advocate for a human-centered approach to AI, emphasizing that success would depend not just on technological capabilities but on thoughtful integration with human workflows and decision processes.

Outcome and impact

As this represents in-progress (as of mid-2025) work, the full impact is still developing. However, several initial outcomes are evident:

• Strategic framework establishment: I created a comprehensive framework that provides a common language and evaluation criteria for AI integration across products.
• Philosophical alignment: I shifted internal discourse from viewing AI merely as a feature set to understanding it as a collaborative system requiring thoughtful design.
• Practical implementation guides: I developed specific guidelines for implementing agentic AI with appropriate human checkpoints, safeguards, and performance metrics.
• Stakeholder engagement: I generated interest and support from key stakeholders, as evidenced by the positive response to my outreach efforts.

The framework is positioned to guide Epsilon's approach to AI integration in ways that maintain human expertise at its core, while leveraging AI's capabilities for more powerful, adaptive, and contextually aware marketing systems.

Reflection

This initiative demonstrates the importance of bringing design thinking and human-centered perspectives to AI implementation. Several key insights have emerged from my work:

• Beyond features to systems: Success with AI requires thinking beyond individual features to consider entire systems of interaction and decision-making.
• Balancing autonomy and control: Finding the right balance between AI autonomy and human control requires careful consideration of when decisions should be delegated versus when they require human judgment.
• Evolution beyond prompts: The current fixation on chat-based user interfaces and prompt engineering represents a transitional phase rather than an endpoint in AI interaction design.
• Meta-design as leadership: Taking a meta-design approach—designing the conditions and parameters under which AI systems create, suggest, and evaluate—represents an important leadership stance in the age of AI.

As this work continues to evolve, it highlights how design leadership can shape technological implementation in ways that amplify, rather than replace, human expertise, ensuring that AI serves human values and aspirations instead of obscuring or displacing them.