Using AI/NLP to enhance your product's user experience

You and your Product team have already built a great and successful product for your customers. Congratulations! As you look into the future, you might ask yourself: what is the NEXT value stream we would like to enable for our users? If the answer is somewhere in between the boundaries of the user experience, then read on, as this might be for you.

In my last article called Digital Transformation: The Importance of balancing Agility, Change and Risk, we explored the intricate balance required to drive a successful digital transformation initiative, with a key takeaway being the need of investing in the right types of technologies that support your business' strategic goals, while enabling a solid base for the future.

Today I want to explore with you one of such emerging technologies: AI, and more specifically how NLP (Natural Language Processing) based tools can be leveraged at the user experience level to complement - if not build brand new - products. Now, to be clear, NLP has been around for many years at this point, but it has definitely gained critical mass toward "mainstream" awareness as of the public release of ChatGPT during late 2022.

As recognized leader Eric Schmidt said, "AI will be the new electricity." Businesses that adopt AI early and effectively will have a major competitive advantage over those that don't. But how do you choose the right AI/NLP "flavor" to complement your product and create a powerful market differentiator for your company? It's like choosing your favorite topping for your favorite ice cream - the right combination will just make everything better!

The 3 Big Buckets

To start, let's look at the three categories that we can use to aggregate the hundreds of AI/NLP based tools that have emerged so far: chatbots, companions, and drivers.

1. Chatbots enable user/machine interaction through natural conversation. They have been around for a while, but have gotten significantly better in recent years and can potentially replace a good chunk of human interaction in multiple areas, such as sales, customer service and even technical support. Depending on your specific needs, there will be various options already you can evaluate: you can develop and train your own model and retain all IP internal to your application, or you can use an API to communicate with popular models like GPT-3 (or soon GPT-4) and expose your content base to the application so it knows how to interact with your product and customers within the limits of your solution.
2. Companions (or copilots) create an enhanced user experience by suggesting the immediate next input based on the previous user actions, but keeping the control on the user's side. Think about the old Office Clip, or "Clippit" that appeared in Microsoft Office from 1997 to 2007, but this time on steroids. Clippit used a relatively basic form of NLP that did not involve the use of machine learning or other advanced AI techniques, but instead used predefined rules and heuristics to provide context-sensitive help to users. Jumping back into today's capabilities, AI-based companion tools can understand context and meaning and therefore be very useful in their suggested interactions. You can find them everywhere nowadays, from word processors for writers suggesting possible content, to messaging applications for the everyday user that suggest the next word or phrase in your text message.
3. Drivers on the other hand, will take text input from the user and assume control of the output. Usually, this type of tool will be used to help execute an action that is complex and/or large by nature, and would require lots of time or skills to achieve it. They often combine NLP with other ML techniques to generate their outputs. Excellent examples of this type of tool are the now very popular text-to-image generators such as Dall-E 2, Photosonic, NightCafe, etc. For example, the cover image of this post was generated using Dall-E 2 using the title of the article with some additional parameters. It then uses NLP to understand the text and GAN (generative adversarial networks) to produce the final image.

Early Adoption

Depending on your product, the potential use cases of those integrations with AI/NLP will vary significantly, but to truly create a powerful market differentiator out of them the key is doing it early and doing it right. After all, copycats will always eventually appear for a successful business model, but if you got there earlier than everyone else, you might have good odds on enjoying that leadership position for longer (until you innovate again, and the cycle continues!)

The best contemporary example for early adoption of this AI/NLP explosion is Microsoft's partnership with OpenAI, integrating ChatGPT and GPT-3.5 learnings into their own model (codename Prometheus) and making it accessible through their Bing search engine, inevitably igniting the search engine wars again. The long-lived King Google is now playing catch up with Bard, but could potentially see its status challenged - at the very least - for millions of users around the world, just for being late to the party.

In conclusion, AI/NLP technology has the potential to create a powerful market differentiator for businesses that adopt it early and effectively to improve their user experience. By understanding the different categories of available AI/NLP tools and experimenting with new ideas, companies can create innovative solutions that differentiate their products and services from the competition.