A strategic framework for generative AI opportunities

Generative AI has hit the business world like a meteor in recent months.

We've never seen a technology rise to the peak of the hype cycle so rapidly. Tools like ChatGPT, with close to the human-level text and visual creation capabilities, have both thrilled and scared us.

To make sense of it all, I want to provide a simple but nuanced framework for thinking about the opportunities that Generative AI (G-AI) opens up for companies. How to tap into these opportunities depends on your specific market, internal G-AI maturity, and resources.

We can categorize the strategic opportunities of G-AI from two broad perspectives:

• internal - how a company uses G-AI to improve internal workflows

• external - how companies leverage G-AI for customer-facing products and services

Within those perspectives, there are roughly two ways of gaining access to G-AI technologies:

• buying access to APIs from leading G-AI product companies like Google or Intercom or

• building your own proprietary G-AI tools or products with models like LLMs (Large Language Models).

Let's start at the most basic level - the starter zone - by purchasing G-AI tools for internal use.

An early framework for analysing different strategic opportunities in generative AI.

1. Starter - buying G-AI tools to improve internal efficiency

Many companies and early adopters have already begun utilizing the "starter zone," where off-the-shelf G-AI tools like ChatGPT are used to increase efficiency and productivity.

Some examples of how G-AI can be used include:

• drafting, iterating, and summarizing documents

• taking notes and summarizing action points from meetings

• generating product descriptions and improving website SEO

• brainstorming ideas in a workshop

• using G-AI as a personal mentor for faster learning
  

Although this is called the "starter zone," it is not necessarily easy to achieve. Our human-powered working habits are deeply ingrained, and it may take years for companies and individuals to embrace collaboration with G-AI daily fully.

To make the most of the starter zone, consider the following:

• providing employees with foundational training on G-AI opportunities

• allowing employees to purchase premium versions of tools like ChatGPT on company cards

• making the most of G-AI that are built into tools people already use - from Notion AI to Microsoft Office

• empowering early adopters to experiment with G-AI and share their findings

• the potential confidentiality and copyright risks of using general G-AI tools

Using ChatGPT for internal brainstorming to launch a product

2. Enhancer - enhancing external products with G-AI

Once you’ve gotten your feet wet in the Starter Zone, consider how you could buy off-the-shelf G-AI to enhance your external facing products and services.

This field is rapidly moving, so expect to see many more API’s open up in the coming months that enable companies to build G-AI powered products with relative ease.

A concrete example is using tools like Intercom to bring G-AI-powered chatbots to customer service. Intercom’s new Fin chatbot is based on the foundational training of the GPT-4 model and supplemented with the help center knowledge of the company using it.

Another interesting development has been the launch of ChatGPT plug-ins. Still in beta, it enables companies like Zapier and Opentable to connect directly to ChatGPT, enabling the chatbot to do the users bidding with these 3rd party services independently.

Intercom’s Fin is a customer service chatbot based on the foundational training of GPT-4. On top of that, it learns from a company’s help center to answer customer questions.

3. Optimizer - building internal G-AI tools

The optimizer builder zone is where it gets more advanced. This zone means building your own proprietary G-AI tools for primarily internal use.

Examples of where this can make sense include:

• creating an image-generator trained on your company’s brand images or illustration style

• building an internal chatbot to support field agents trained on the specific domain knowledge and internal materials of your company

• using a custom internal chatbot to provide instant insights based on all of your internal confidential materials

The line between buying and building is getting fuzzy as unless you’re Microsoft or Google; the odds are that you will be building your G-AI solutions on top of cloud-based computing power offered by companies like NVIDIA.

This makes building G-AI more accessible as you don’t have to invest in the massive computing infrastructure required to teach the data-hungry deep learning models.

For example, last week, a Goldman Sachs executive hinted it might build its own internal LLM-based chatbot for its over 45,000 employees to tap into their internal knowledge.

The Optimizer zone is particularly relevant for companies that:

• have massive amounts of proprietary knowledge trapped in silos like millions of emails and documents (like Goldman Sachs)

• sit on top of an IP gold-mine of content like images, text, and

• make decisions based on extremely sensitive information like finance or health

• rely on making better decisions and or content for their competitive advantage - like trading or media

Companies are starting to build their own internal generative AI tools to tap into the company's collective wisdom combined with all of the accessible knowledge online.

4. Magician- building external G-AI driven innovative products

This is the ultimate level of utilizing generative AI (G-AI): building external products and services that rely on G-AI to create value.

Many startups without the baggage of a legacy business are jumping directly into this zone. In the coming years, several industries will be upended by G-AI-powered newcomers that tear entirely apart the conventional way of solving customers' problems.

Imagine a travel assistant that handles all your travel-related questions and tasks, gives recommendations, makes all the bookings, and pings you when your boat is leaving. All through natural conversation via text or voice. No need to close pop-ups on websites or stay on hold for customer service.

The most striking first example of this is perhaps the online search industry. ChatGPT has become the first destination for me and millions of others to ask for knowledge that’s not tied to current events. The experience is superior to browsing separate websites to find or combine the correct information.

No wonder Google has accelerated the launch of its own G-AI companion Bard.

Zooming out for a macro view, our dominant way of talking to computers will likely shift from the graphical interface to natural language.

We get far better answers faster by interacting with machines like we do with humans - by talking and writing. It’s not difficult to see how disruptive this is for companies that have built their business on a “traditional” online model - let alone an offline one.

To start, build a solid foundation

Hopefully, this simple framework for mapping the opportunities of generative AI has been helpful.

To start, it's helpful to build a solid foundation of understanding and exploration of G-AI in the Starter zone. Once the foundation is in place, it's worth analyzing the particular characteristics of your strategy, customers, market, and resources to determine what bets you should place in the other three zones.

Lastly, consider moving at a relatively fast pace but still keeping a firm grip on the ethics of AI. As this tool is potentially more powerful than anything humans have ever created, the enormous opportunities are balanced by gigantic risks.

Business leaders should view the opportunities of generative AI through at least four lenses: their impact on the bottom line, their customers, their employees, and society. But that’s a topic for another future deep dive.