)
)
)
Courses
Conversation design has evolved from a niche specialty to a critical enterprise function. As organizations rush to implement large language models (LLMs), the role of conversation designers has become more crucial than ever. But how do you build and advocate for a conversation design practice in a corporate environment?
The conversation design landscape has transformed dramatically over the past few years. While traditional conversation design focused on crafting specific dialogues and responses, today's practice encompasses a broad skill set including prompt engineering, UX design, interaction design, and more.
Modern conversation designers have become “full-stack” practitioners who are often required to wear multiple hats. But the list of things that they need to master is growing: from information architecture to flow design, to intent training and classification, prompt engineering and LLM evaluation. It is one of the reasons why CDI has been advocating for teams with specialists rather than a team of generalists.
A successful conversation design practice in the enterprise rests on four fundamental pillars:
Guidelines and standards: establishing clear standards for prompt design, design patterns is essential for maintaining quality and consistency. Well-documented standards serve as educational resources for stakeholders and create alignment across teams, reducing confusion and improving collaboration.
Design deliverables: modern conversation design deliverables have evolved far beyond simple dialogue flows. Today's comprehensive documentation includes detailed prompt libraries and engineering specifications that guide LLM implementation.
Stakeholder management: regular touchpoints with stakeholders are essential for creating buy-in and demonstrating value. These collaborative sessions educate people about the importance of conversation design while gathering valuable insights from different parts of the organization.
Advocacy and education: positioning conversation design as a strategic function requires consistent effort and clear communication. This involves creating educational content, delivering presentations, and demonstrating the value of conversation design in achieving business goals.
To successfully implement conversation design in an enterprise setting, you also need to ensure close collaboration with data science and engineering teams. Working systematically reduces technical debt and improves maintenance efficiency. When design patterns are consistent, updates and optimizations can be rolled out more effectively across the entire system.
Understanding LLM capabilities and limitations is crucial for modern conversation designers. LLMs excel at classification tasks, content summarization, and taking complex information and distilling it into digestible content. They’re particularly strong at format transformation, converting content between different styles and structures.
However, generative AI comes with important limitations that must be carefully considered. LLMs struggle with non-public information, making them less reliable for private or proprietary data. Brevity can be a challenge, as LLMs tend to be verbose and struggle to provide concise responses. The lack of accuracy can cause problems, particularly when dealing with specific facts or figures, requiring verification or a human in the loop.
As enterprises continue to adopt generative AI and conversational AI, the role of conversation design becomes critical in managing the balance between technology and user experience. As user advocates, conversation designers should have a seat at the table when organizations plan for future initiatives. By building a strong Conversation Design Practice founded on proven principles, maintaining high standards, and consistently demonstrating business value, conversation designers can lead their organizations toward more effective and meaningful AI interactions.
Do you want to know where your practice stands today? Start by assessing your maturity through our free Conversational AI Maturity Assessment.
