The Future of Generative AI in the Enterprise

The world is abuzz about large language models and the promise of AI. But how will all of this play out within the enterprise? We talked to the head of AI research at one tech company to find out.

👋 Hi and welcome to The DX Report — all about Digital Transformation, the Digital Experience, and the Digital Enterprise. I’m industry analyst, author, and speaker Charles Araujo, and I’m all about providing insights and analysis for enterprise IT leaders as you make the big bets about your organization’s future!

---

The Bet: Incorporating Generative AI Into Your Broader AI Strategy

It’s been pretty hard to miss the crazy buzz around large language models (LLMs) and generative AI in the enterprise tech space.

Despite the fact that we’re already seeing the hype cycle kick in with ChatGPT’s traffic falling for the first time, it has unquestionably kicked off a mad rush among enterprise tech vendors to not be the only ones left standing outside the big Generative AI tent.

But this rush to incorporate generative AI at all costs has left many enterprise IT leaders trying to separate hype from real opportunities to leverage it and make a strategic difference in their organizations.

As the Wall Street Journal reported:

Businesses are facing an influx of new artificial-intelligence tools, many of which overlap and cause confusion for employees, as corporate-technology sellers race to capitalize on the generative AI trend.

This situation creates a somewhat bizarre bet. There’s little question that generative AI is going to enter the walls of the enterprise in some way — it’s being baked into almost everything in some fashion. So, the real bet is how much attention you should pay to it, now and in the future, and in what areas.

The Briefing

The term artificial intelligence (AI) was already fraught with confusion. It’s a catch-all term that, as a result, doesn’t mean much of anything.

While Generative AI is a little better in that it at least refers to a specific application of AI, how an organization or tech company applies it varies wildly. So, like the broader term, the mere statement that a tech company is incorporating generative AI doesn’t tell you much.

But as you evaluate your approach to generative AI, there are really three primary factors you should consider: corpus, context, and use case.

The Corpus Starting Point

If you’ve been following the evolution of this space, you’re probably starting to hear some chatter about the importance of an LLM’s corpus.

The corpus is critical, because it is the foundation on which a provider builds its LLM. As FastCompany explained in its recent primer on the subject:

When used in the artificial intelligence realm, the term “corpus”…refers to the metaphorical “body,” or collection, of data that was used to train the AI. This corpus is the material the AI reviews to become intelligent in whatever it was designed for.

Every AI’s corpus will be different, because it is humans who decide what kind of data they want to train an AI on. And the corpus the humans decide to train the AI on will depend on what they want the AI to be proficient in.

As ChatGPT burst onto the scene, the corpus didn’t matter much as it was the only real game in town. But with several competing LLMs emerging and a bevy of heavyweights promising to build their own, understanding the corpus will be essential in determining its applicability and potential effectiveness to your particular use case.

Most importantly, as Zoho’s Director of AI Research, Ramprakash Ramamoorthy, explained during our recent interview (see below), you will likely need the flexibility of accessing multiple LLMs, all with different corpora (what an amazing plural, right?), to meet the range of business use cases you’ll eventually want to tackle.

Creating Meaningful Business Context by Integrating AI Approaches, and Public and Private Data Sets

The second major consideration when it comes to how you apply generative AI is your ability to use it in context.

The most common uses of ChatGPT, or any of its newer competitors, doesn’t provide any context. You chat with the core model and get what you get. Microsoft’s Bing version and the new ChatGPT web plugin allow you to point the LLM at live search data, but that’s still pretty generic.

As Zoho’s Ramamoorthy explained in our interview, you only release the true power of generative AI when you combine the power of LLMs with two things: other approaches to AI, and targeted public data.

In Zoho’s case, that means combining its newly released, ChatGPT-powered generative AI functions with its homegrown AI tool, Zia. Zia has access to all a company’s internal data stored in Zoho’s suite of CRM, marketing, accounting, HR, customer support, and other tools. It is then able to combine that intelligence with summarization and generative capabilities from ChatGPT (without sending any data out of a customer’s environment. Ramamoorthy covers Zoho’s approach to privacy in some detail).

Likewise, he shared examples of combining targeted public data, for example a competitor’s publicly available sales figures, with your private data and then leveraging LLMs to answer questions about competitive positioning.

The key in understanding when and how you may be able to leverage LLMs is in understanding your ability to create this sort of context via various AI tools and a mix of data.

Focusing on the Use Case

The final “aha” moment for me in my interview with Ramamoorthy was at the end when I asked him about how enterprise leaders should be looking to the future.

“Process optimization and revenue maximization will be the goals of AI for any big enterprise,” he said.

I actually think there are more use cases than merely process optimization and revenue maximation that will create value for enterprise leaders (its use in strategic planning, for instance, could be very powerful), but the underlying point is critical: the effectiveness of an LLM in the enteprise will be in direct correlation to your effectiveness in applying it to the right business use case.

Of course, that sounds obvious. But much of the early talk around LLMs has centered around fairly transactional use cases. Those uses will likely drive some efficiency gains and bragging rights (for a short while).

However, the real game-changing uses of generative AI will emerge as organizations focus on how they can apply generative AI deep within core business processes, revenue generating engagements, strategic planning activities, and other places where its impact will be profound.

My interview with RamamoorthyI goes into much more detail and provides great insight into both how Zoho is approaching this evolution, and what it may mean to you as you seek to leverage it.

The Brass Tacks: Invest, Pass, or Hold?

🚀 Invest — and Hold?

This is a tough one for me because there’s no simple answer and no single bet.

As I mentioned, generative AI is coming into your house no matter what — and if you aren’t already, you’ll be fielding mountains of questions about leveraging it.

So, you really don’t have a choice but to invest. Being a naysayer on the sidelines is just going to get you a one-way ticket to the stands.

At the same time, some patience, skepticsm, and due diligence is prudent.

I think companies like Zoho, BigPanda (which recently introduced a native generative AI interface to its tool), and Automation Anywhere (it is leveraging it to, amonth other things, allow business users to extract information from across multiple systems) are doing some great stuff.

Still, it only matters if it matters to you. You can easily get caught up chasing every new evolution of this tech — and that would be a bad bet.

Use the three factors of corpus, context, and use case to place targeted bets on where you apply generative AI in meaningful ways. But when you place those bets, beg big.

The only way you’re going to make a significant difference with this technology is if you go heavy and get deep. Otherwise, it will just be surface fluff. And if you can’t go deep or don’t trust the bet, don’t make it.

So, that’s the brass tacks for my point of view, but what do you think? Agree? Think I’m completely off? Let me know!

And don’t keep this conversation to yourself. Invite your friends and associates to weigh in!

Disclosure: Zoho is a client and paid for my travel to its Zoholics event, where I conducted the above interview. However, Zoho had no editorial input into or review of this content. BigPanda is also a client.