How to Build AI-First Businesses in the Age of Artificial Intelligence

Imagine a world where training your fluffy friend is as easy as gossiping with your next-door neighbor; where whispers of "good boy" or "who's a good girl" are met with more than a wagging tail or a contented purr.  

Sounds wild, right?  

But hold onto your hats because that future is closer than you think! As the lines between science fiction and reality blur, faster than anyone imagined, business leaders are struggling to make sense of this rapid change while worrying about meeting these new sets of demands from their customers.  

At our very first flagship event CXUnifiers, Allie K. Miller, top AI leader, advisor and investor, gave a complete rundown of how AI is disrupting businesses today. And how business leaders should plan their AI investment strategy.  

In this blog, we’re going to highlight the key points from Miller’s talk and share her methodology for building AI-first businesses, in addition to helping you understand the framework for investing in AI use cases.  

4 lenses to understand change in the dawn of AI 

Heraclitus, a Greek philosopher, said that the only constant in life is change. AI has accelerated not just change but the rate of that change too. Your businesses were exposed to changes before, but this change is a new beast.  

“No matter how big the change is, AI is a tool that can handle this rate of change – Allie K Miller 

Miller used four parameters to put this into perspective.

1. Scale

Scale is the first lens. How has scale changed in the age of AI? Miller drew attention to the products that hit 100 million users the fastest as a means to understand scale. For instance, it took seven years for the Internet to hit 100 million users. Gen Z-friendly social media apps like WeChat and TikTok took a year. However, the products occupying the top two positions on the leaderboard are AI-based – ChatGPT and Character AI.  

Any guesses on how much time they took to hit 100 million users?  

Less than 6 months! 

AI products are competing in a league of their own.

2. Performance

The second lens is performance – comparing AI’s performance versus that of humans.  

Look at the graph below that displays the time it took AI to outperform humans. It took 16 years (1998 to 2014) to outperform humans for a task like handwriting recognition. Compare this with image recognition which took only six years to outperform humans for a task like image recognition. From 2009 to 2022, AI models became progressively smarter, and it didn’t take more than a year for these models to master tasks like reading comprehension, language understanding and code generation.

When it comes to enterprise business change, it took four years for AI to go from sub-human performance to super-human performance on Massive Multitask Language Understanding (MMLU). This is faster than real-estate business planning. 

Massive Multitask Language Understanding (MMLU) is a groundbreaking approach in the field of natural language processing (NLP) and AI that aims to enhance how machines interpret and process human language significantly. By integrating the concept of multitasking — where a single model is trained to perform various tasks — MMLU represents a leap toward creating AI systems that can understand and interact with human language more nuanced and comprehensively.

3. Cost 

Cost is a huge factor. This wouldn’t matter much if only big companies got to play with these tools. AI models, despite getting bigger, are also getting more cost-efficient, leveling the playing field for everyone.

“The cost of GPT 3.5 or GPT 4 or Claude or Lama 3 are all going to plummet in the next few years – Allie K. Miller 

4. Accessibility 

The last lens is accessibility. Before the launch of GPT 2, AI was available only to machine learning engineers and data scientists, who are roughly one million around the globe. Then in 2019 when GPT 2 was unveiled, around 28 million developers were able to play with it. Today massive AI models are available to five billion internet users, and they have much more leeway than those developers. For instance, prompting can now be done in the user’s native language.

If you have banned AI and/or are thinking of restricting its use in your organization, think again.  

Frankly, there’s a lot of change happening right now – scale, performance, cost, and accessibility.  

Have you thought about how to pivot your business to tackle and maybe leverage this change to your advantage?  

The rise of AI-first business models

Traditionally businesses catered to a locality or a community of people. They found it hard to grow beyond these communities. Then came the digital boom and with it we saw the mushrooming of digital native businesses like Netflix, Pinterest, and Airbnb. They were unbelievably efficient at scaling. But with scaling complex problems cropped up and these companies’ adaptability was severely restricted owing to their size. 

Then came the dawn of AI and with it a new set of businesses.  

40,000 AI startups to be precise. 

Miller worked with top AI founders, AI researchers, and AI investors to study AI startups and understand what they are doing differently. She and her team wanted to understand what these businesses were building, and her biggest takeaway was that these companies were neither building AI products, nor infrastructure to support AI initiatives, not even anything remotely related to science and engineering. Instead, they were building a new business model.  

“The big takeaway for my team is that they were not just building AI products, but rebuilding their entire business model from scratch with AI at the forefront of everything they do.” --- Allie K Miller 

Change is inevitable and we are at a pivotal point. No matter how big the change is, AI is a welcome addition to your toolkit that can handle this rate of change as demonstrated by the emerging AI-first businesses. “Generative AI allows you to understand large quantities of unstructured data better in a way that you’ve never been able to do before”, says Ragy Thomas, Co-CEO, Sprinklr. Listen to his insights on how to turn AI into a distinctive advantage for your business.  

But how do you build an AI-first business?  

Miller’s 3P methodology for building an AI-first business

According to Miller, there are three levers: People, process, and products.  

People: How to supercharge your employees

People are the crux of any business. Supercharging your employees not only means improving their productivity but how you make your employees happier. In other words, how you liberate them from their routine and tedious tasks.  

Let’s look at an example of how AI is employed to assist wealth managers.  

Every single wealth manager in Morgan Stanley has an answer bot. The bot has access to 100s of thousands of pages of proprietary data. 300 front-line workers can leverage the most up-to-date info. Whenever a customer asks an unfamiliar question (for example, how can I set up my Morgan Stanley account?), the wealth manager heads over to their fancy bot, gets the answer, and brings it back. What used to take a few minutes to several hours is now reduced to seconds, saving both customers and wealth managers' time.  

Morgan Stanley understood that maintaining customer relationships is the key that is driving their business, and that their customers may not be as educated about AI as their employees.  

With the answer bots’ assistance, wealth managers are now spending more time building relationships and less time searching for answers.  

Process: How to supercharge your operations

Here you need to think how AI can power your back-office operations like gathering data and deriving insights from that data, team alignment, system-to-system communication, and interoperability.  

For example, Walmart experimented with a negotiation bot that can talk about contract terms, negotiate, and get deals signed from their long-tail vendors. Long-tail vendors are small vendors that do not have a dedicated Walmart staff assigned to them. They waited for more than two weeks to get a response to their emails. Walmart launched this bot in the hopes of closing 20 percent of deals from these vendors.  

Can you guess how many deals Walmart closed?  

A whopping 70% with a cost saving of 3%.  

That’s not all, wait times for receiving an email response reduced significantly and more than 90% of vendors were happy to continue their engagements with the bot.  

Products: How to supercharge your solutions

This is customer-facing and deals with ever-changing market demands. It is a high-risk category because it drives revenue growth, not just cost savings.  

Let us look at an example to understand how to build your solutions with AI. Stitch Fix, an online personal styling service for clothing items, used AI to mass generate their product descriptions. They looked at product reviews from their customers and used it as a cue to generate product descriptions. For instance, if a product description had the word “wedding dress” and all the user comments said that the outfit is better suited for work and not for a wedding party, then Stitch Fix used these comments as inputs to their AI model to fine tune the description for that product. They had their expert copywriters who were trained on their brand and voice to review and edit these product descriptions. 

Stitch Fix can also generate outfit combinations using AI. They generate 13 million outfits a day and 43 million outfit combinations a day. And the reason this scale matters is because this is the only way to get to hyper-personalization.  

Here are a couple more use cases for each of the Ps in Miller’s 3P model. 

How to start your AI investment journey

Miller shares a list of questions to help you with your AI investment journey.  
 
1. Is your use case a core part of your business? Or is it stretched further out?  

The more crucial it is to your business, the more you want to own that solution. For example, Audi built their own AI models to generate tire designs. But if they needed a solution to a customer support problem, then they would turn to an external or a third-party vendor for help.  

2. What is your AI investment timeline?  

If you are newer to AI, you should not take on anything that is in the years category. The time- to- value is about six months for most things like personalized recommendations. You want to make sure that the investment timeline is short in the beginning so you can gain momentum. Some of the other questions that you need to ask are what is your R&D/ pilot budget? Do you have the expertise in-house, or do you have to source it outside? What requirements need to be addressed from a compliance, privacy, and governance angle? And then, of course, you want to consider who the end user is to figure out how you're going to design this solution. Based on your answers to these questions there are five ways to deploy AI.  

1. BYO (Build Your Own) model: This is building your model from scratch. If your use case is not something that is core to your business and you don’t have a sizeable budget and longer investment timeline, then this may not be a good fit for your business.   
2. Fine-tune model: Stitch Fix used this model. They took a base model and customized it for a specific task.  
3. Quering model: This is going to be a quite common model. This is where you spin up your internal Google, where you can ask questions about your internal documents and bring it back.  
4. API (Application Programming Interface) and subscribe model: You either use an API or subscribe to a third-party SaaS (Software as a Service). For example, customer support use cases fall under this model.  

Conclusion

In conclusion, we're witnessing the dawn of a new AI era, marked by unprecedented change. AI isn't just restricted to specific use cases; it's reshaping entire business models.  

“The big change that we saw was that there was a business model paradigm shift.” – Allie K. Miller 

From scaling and performance to cost and accessibility, AI is transforming industries. Examples shared in this article, from customer support to product descriptions, highlight AI's profound impact across business verticals. But human expertise remains vital. Therefore, collaboration and synergy between humans and machines is key. Be sure to remember the three Ps: people, process, products — empower employees, streamline operations, and innovate your products — to stay agile in this fast-changing landscape.  

Discover interesting stories of AI in CX by visiting AI-Wise, our curated content library aimed at boosting your CX and helping you achieve radical productivity using AI.