What is Natural Language Experience Design
Think about a great conversation you have had recently with your friend, doctor or work colleague. The kind of conversation that felt easy, where you didn’t overthink your response or the questions they asked, and you left feeling understood rather than frustrated like the woman in this meme.
What did all of these great conversations have in common? The conversation flowed effortlessly and you left feeling understood and satisfied with the outcome.
Great conversations feel natural. But they’re anything but simple.
Everyday human conversations follow countless invisible rules and heuristics. Through a lifetime of social conditioning, we have learned to intuitively manage turn-taking, interpret tone, infer meaning from context, and respond appropriately depending on who we are speaking with.
Recently, I have been thinking a lot about these invisible mechanics that govern human conversation and how they should inform the design of conversational AI products. More and more of our “conversations” are not with people at all. Increasingly, we are interacting with conversational AI, which we expect to quickly understand our intent, respond naturally, and help us accomplish tasks reliably.
This is where Natural Language Experience (NLX) Design comes in.
Natural Language Experience design is the practice of designing human-AI conversations in a way that feel natural, intuitive, and helpful for users. NLX does this by embedding the same implicit conversational mechanics that make human-human conversations work so well.
Why NLX Matters
We have all seen conversational AI solutions fail in ways that feel avoidable in hindsight. For example, DPD previously made headlines when its AI agent swore at users and criticised the company. This was not a technical AI failure, it was an NLX (and Responsible AI) design failure. Thoughtful NLX design could have easily prevented a very public breakdown of a publicly facing AI agent and saved DPD from some embarrassment.
NLX design isn’t just an abstract design theory. It is a practical solution to the challenges that the biggest organisations in the world are facing right now when deploying agentic AI solutions. According to PwC’s recent AI Agent Survey, 88% of the world’s top companies plan to increase their investment in agentic AI over the next 12 months. Deploying enterprise agentic AI solutions comes with challenges though. Results from the same survey indicate that the key challenges faced in fully realising value from agentic AI, include:
- a lack of user trust in AI
- difficulty driving adoption among employees
These are fundamentally human challenges which stem from how people perceive and trust AI. Thoughtful NLX design can make a meaningful difference in resolving these challenges by designing conversational interfaces that increase user trust and long term adoption of conversational AI solutions.
Where Do I Start? – 4 Foundational Pillars of NLX
By now, you’re probably wondering what it takes to design truly great natural language experience. It’s the billion-dollar question everyone is racing to answer.
As with any complex challenge, there is no single simple solution. NLX design requires a multidisciplinary approach. In my view, NLX sits at the intersection of four foundational pillars:
Note: The key word here is intersection. While some concepts overlap across disciplines, each pillar offers its own distinct perspective. Bringing them together is what enables a truly impactful NLX design. I’ll dive into each pillar in more detail in upcoming articles, but here’s a quick overview to get us started.
1. Traditional UX Design The fundamentals of traditional UX design such as usability, accessibility, user control, and error handling still apply to conversational AI interfaces. NLX draws from traditional UX, applying its principles to conversations, intent handling, and the behaviour of AI systems.
2. Responsible AI (RAI) Trust and transparency are essential in any high-stakes conversation, especially when one side is AI. RAI principles like transparency and explainability (XAI) shape how users understand, adopt and build trust with conversational AI products.
3. Linguistics Language is the interface we are working with, so we need a solid understanding of how language is constructed. Designing for semantics, syntax, discourse structure, and tone ensures the conversational AI product interprets user intent accurately and responds in ways that feel natural rather than clunky.
4. Human Psychology To create intuitive natural language experiences, we need to understand how people think, decide, and communicate. Elements like cognitive load, heuristics, decision framing, nudges, and micro-interactions all influence how users engage with conversational AI products.
We have established in this article that NLX design is absolutely central to the successful user adoption of a conversational AI product. In my opinion, NLX design creates successful and fulfilling human-AI interactions by combining elements of traditional UX, Responsible AI, linguistics, and human psychology.
Over the coming weeks, I’ll break down key considerations for each of the four foundational pillars of NLX. By the end of this series, you’ll see how a multidisciplinary NLX approach can transform human-AI interactions from frustrating and unfulfilling to natural and trustworthy. Stay tuned!
