Cutting Through Generative AI Hype: Use Cases And Drivers For AI-Driven CX
Vin Vashishta
AI Advisor | Author “From Data To Profit” | Course Instructor (Data & AI Strategy, Product Management, Leadership)
It’s getting harder to filter noise out of the Generative AI conversation. Generative tools will change customer service and reset customer expectations. The hype cycle has us looking too far out into the future and missing opportunities that are right in front of us today.
This week, I was part of a panel discussion about how Generative AI is reshaping customer perceptions and expectations. Businesses I advise are working on early implementations. Those solutions won’t be here in time for the US holiday season, and I wouldn’t risk launching untested technology right now, even if they were finished. I expect most first-generation Generative AI solutions will roll out early next year.
I also expect many businesses to be caught off guard by the customer response. ChatGPT’s adoption curve showed us how quickly Generative Interfaces and experiences take off. There will be Generative AI “haves” and “have nots.” Customers won’t wait for the “have nots” to catch up.
Businesses need a Generative AI strategy that spells out likely changes in customer preferences and where the business has opportunities or threats to consider. It should consider multiple perspectives. SAP recently held its CX event and shared its vision of how businesses will leverage Generative AI to serve customers better. The company has implemented Joule, and the Generative Assistant is a powerful example of what’s possible.
Here’s my take. It’s based on early products and trials I have worked on or participated in. It's hard to imagine what's about to happen until you see how people respond to Generative Interfaces and productivity tools.
New Behaviors Are Emerging With Generative Interfaces And Tools
When generative interfaces are designed without a lot of friction between customers and the tools, people integrate them into their workflows rapidly. There’s very little training required. Well-designed generative interfaces are habit-forming.
People use them for a few weeks and either abandon them or never want to do their job without them. The tools that gain traction are more reliable than those that people leave behind. If the generative tool makes too many mistakes or is too generic, users abandon it at very high rates.
However, users get hooked on the new way of working with a reliable tool, and that’s where habits quickly take hold. Take the tool away, and workers are less productive but also less satisfied. People get frustrated with losing access and being forced to perform tasks the old way.
That’s why I believe customer preferences will change quickly and irreversibly. It’s hard to believe until you see someone get angry over losing access to a generative tool. Customers will form similar habits and won’t tolerate businesses that provide legacy experiences or services.
Generative tools seem to make people more collaborative, something I observed with generative coding tools. One of the greatest crimes you can perpetrate against a software engineer is interrupting their coding flow. It’s different when they are using a generative coding assistant.
Instead of writing all their code from scratch, most developers use the generative coding assistant to build templates and fill in the gaps. What’s interesting is they’re able to context switch faster. Answering emails or breaking to attend a meeting isn’t as disruptive.
My hypothesis is that software engineers have a different type of flow with the generative coding assistant. Since they already collaborate with the generative tool, breaking that flow to work with someone else doesn’t involve as much executive function. There’s less fear of losing their train of thought.
I believe this will translate well into customer service and sales experiences. Customers will open up to generative tools and interfaces in ways they might not with a salesperson, support technician, or digital customer experience. Customers will be more open to working with salespeople as part of the process rather than viewing them as an unwanted interruption.
Putting The Pieces In Place Early
Based on these trends, I see a few simple use cases that businesses can implement in the next 6-12 months. There are key enablers that must be in place before a business can move forward with any of these types of use cases:
Customer and product data from across the business must be accessible to LLMs.
Data must have customer context, the metadata about what customer activity, experience, or process generated the data.
Tools and infrastructure must be in place to support data curation, LLM training or fine-tuning, prompt engineering, and production deployments.
领英推荐
There’s not enough time to build all the pieces internally. Business leaders must look to third-party solutions for support with the technology but also the transformation. These are massive investments, but the alternative is slower and more expensive. Inefficient transformation isn’t an option.
OpenAI is partnered with Bain to deliver professional services that help customers with use case selection, transformation, and technical implementation. SAP has an enterprise-wide platform solution, partners, and consultants to support every phase. Most companies do more than drop technology on their customers’ doorsteps.
Technology and competitors are moving too quickly for early-maturity companies to learn through trial and error. These pieces must be in place so businesses can leverage Generative AI for more than competitive maintenance. Mistakes and false starts are more expensive than buying third-party support.
Car Buying Experience – Bringing Digital Experiences Into The Real World
CX often focuses on bringing real-world experiences into the digital space, but Generative AI enables the opposite. Customers famously hate the pressure and experience they get on car lots. Many have turned to online car buying to avoid the hassles they experience at the dealership. Generative Agents support bringing digital experiences to the physical shopping experience.
A smartphone app that supports text, images, and voice to answer buyers’ questions is a simple way to bridge the desire to see and touch a car with the online shopping experience. Customers are uncomfortable asking all their questions during interactions with car salespeople. They have fewer inhibitions about asking questions in the digital arena.
A Generative Agent, finetuned on car product specifications and customer reviews, can answer a range of questions. Customers get a see, touch, and potentially drive physical experience with the low pressure of an online sales process. The salesperson gets involved when the customer is ready for a test drive or has questions the app can’t answer.
This reverses the journey, and customers ask for the salesperson instead of the salesperson initiating the conversation.
Repair Shop – Scaling Customer Service With Intelligent Solutions
Auto repair shops have a different problem. Customers can’t always get all their questions answered. Repair shops are staffed to complete the work, not explain it. Adding more customer service staff or having mechanics spend more time with customers adds overhead.
The same type of smartphone app can support customers in the repair shop setting. Before they get to the shop, the app can help customers understand the repair process and costs. It can manage the appointment scheduling process, walk customers through check-in, and provide repair status updates. After the appointment, customers can get questions about the repair and next steps answered.
This use case may sound dull compared to what the hype cycle promises, but businesses can realize productivity gains, scale customer service without hiring staff, and improve the customer experience. The finetuning data is accessible even when the business is at an early data maturity phase. These use cases require smaller LLMs, which are easier to finetune, deploy, and support.
The app concept also delivers a competitive advantage over businesses that only provide legacy customer experiences. Many competitors are distracted by shiny object syndrome and missing ways to deliver customer value in shorter timeframes.
?
Generative Interfaces and Agents will support new customer experiences. They will be habit-forming and rapidly reshape customer expectations. Businesses must begin now but don’t need to do all the work alone. Focusing on near-term opportunities builds momentum, domain knowledge, and early returns. What sounds dull compared to the hype is exciting to customers. Early returns support more investment and the long-term roadmap. Transformation is a continuous process and a necessary one to keep up with evolving customer expectations.
?
SAP sponsored this post, so it’s free for everyone. I appreciate their support for insights and deep dives usually hidden behind a paywall.
Partner Alliance Marketing Operations at Data Dynamics
8 个月Insightful post! Your take on Generative AI's impact on customer service is on point. The crucial role of data as an enabler for GenAI deployment is emerging with the rise of adoption of GenAI. Turning data from storage to an asset, understanding its 'why, what, and where,' is essential for successful implementation. Here is an article on how data is acting as a fuel for the next-gen GenAI innovations. https://www.dhirubhai.net/feed/update/urn:li:activity:7146485133213179904
When you mentioned the Open AI drama, i thought you meant the whole sacking issue
Principal @ dss+ | Digital for Sustainability, Risk Management, Operations Excellence
10 个月Absolutely agree! Building a solid foundation and starting with simple, practical use cases will pave the way for successful implementation of Generative Interfaces. ????
Totally agree Vin Vashishta. Confusion and uncertainty does not foster the confidence needed to make long-term investments. The industry was just getting its mind around the "new world order" of AI. Now, we have a new "new world order" to consider in the AI realm. But this time, investments will happen more slowly as software developers wait to see if their would-be generative AI partners have market staying power.
I make you money and create value with data and AI, built multiple billion dollar plus data and AI driven products
10 个月Smaller LLM's work just fine for most businesses and don't cost that much now. I do think OpenAI had too much attention and maybe now people will see there are other options available.