Main content

Next ‘24 - Google Cloud CEO on why an open AI platform with choice, one which allows you to differentiate, is essential

Derek du Preez Profile picture for user ddpreez April 11, 2024
Thomas Kurian, CEO of Google Cloud, explains why buyers should be seeking open AI platforms that allow them to differentiate through customized experiences, and why this is key for future competition.

An image of Thomas Kurian, Google Cloud CEO
(Image sourced via Google Cloud)

Earlier this week, diginomica outlined Google Cloud’s AI strategy, following a series of product announcements at the vendor’s annual Next conference in Las Vegas. CEO Thomas Kurian took to the stage to showcase how Google Cloud has built a platform that allows buyers to choose from a variety of models, whilst fine tuning them and customizing them with their own enterprise data. One of the key elements of the pitch was the idea of ‘AI agents’, which are built through Google Cloud’s Vertex platform, and serve a number of roles, including customer agents, creative agents, employee agents, and data agents. 

What stood out from the demonstrations on stage, and from the conversations we’ve been having with customers this week, is the level of abstraction that AI allows, between the end users and the processes that typically support systems and experiences. For example, if an employee can converse with a system via text or voice, describing what needs to be done, and an AI agent carries out those tasks for you on your behalf, the underlying processes, forms and workflows that typically would make up that experience are somewhat removed from the manual work that has had to be carried out in the past.

Equally, if a customer can converse or chat with an AI agent about what they’d like to buy, or problems that they’re having with a company, which the AI agent then carries out on the customer’s behalf, this is a very different ‘experience’ from clicking through orders, or filling out forms to get what they need. 

With this context, I was curious to understand how Kurian sees this ‘future of experience’ impacting technology purchasing decisions for buyers, particularly as it relates to the backend systems that ‘operate’ how a business runs. System of record vendors are often advocating for an approach that sees Generative AI built into the processes of their systems, which can be utilized by buyers. Whereas it seems that Google Cloud is advocating for more of a ‘build’ approach to generative AI, where companies adopt an AI platform that allows them to have a pick of models, select what data they use, tune and manage those models, as well as build new experiences. 

Commenting on the importance of choice for customers as it relates to the development of AI, rather than relying on a vendor integrating one model into their systems, Kurian said: 

We are the only company that’s vertically optimized and open. Vertically optimized meaning we've been working at the lowest level on AI systems. You've seen many of the new chips that we've introduced. We are building models, with a new version of Gemini 1.5 Pro, as well as a new Imagen model. 

We've got a tools platform, Vertex, which has had many, many advances. And we use Gemini across the portfolio. 

Most importantly though, we've always had a philosophy of ‘open’. And why is that? Because when customers go from a proof of concept to production, and they're using something across their organization, they want to standardize on a platform and have a choice of models. 

It’s just like saying, imagine you went to a grocery store and it only sold you one product, in one category. You could never cook a meal with it. If you were going to just have a proof of concept, that might work. But if you want to deploy it organization-wide, you want all the choices that you would get in the grocery store. And so I will say that is the heart of the difference in what we're doing versus the competition.

It’s clear to me, using this analogy, that Google Cloud is advocating for customers to build their own unique applications and experiences using AI (‘cooking a meal’). Kurian said that most of Google Cloud’s competitors will say that they are ‘vertically integrated’, which he believes just means that they’re a closed system with limited choice. And that’s true to some extent. We are seeing other vendors selecting a model that they believe will win long-term (typically provided by OpenAI) and integrating that into their systems. However, as Kurian noted: 

Just today I met with customers that are using Gemini, Vertex and Anthropic. Gemini and Llama and Vertex. And so we're standardizing customers on our platform, by giving them more choice. We are very optimistic, given the interest we're seeing from customers, given the adoption we're seeing in many industries. 

Every customer who was at this event could have chosen another platform. The reason they're at this event, the reason they're using our technology, is it works. It provides them the right solution. And we have the right people supporting them. 

What customers want is an AI platform, not just a model. Picking one model is not a useful thing when models are changing every other week. What people need is a platform. And a platform that has all the capabilities we've talked about - tuning, augmenting a model, helping them retrieve data, evaluating the quality of the model, testing it, deploying it, monitoring it. We offer a platform, others offer a model.

What this means for the backend

If the future is an ‘open’ platform that provides a certain level of choice for buyers to build their own AI applications and experiences, whilst these experiences also abstract away from the backend processes, what does this mean for buyers who are facing a range of purchasing decisions? 

Enterprise buyers are currently grappling with every vendor telling them that their AI offering is the offering to go for, whilst they attempt to pull together a coherent AI strategy. It’s worth noting that some of this will be decided for them, as some vendors simply integrate generative AI experiences into their existing experiences. 

But it’s also true that some buyers will be thinking about how to go further and differentiate to compete. The future is still uncertain regarding the impact of AI and how the decisions made now could likely have an impact on competition in a year or two. For instance, a buyer that adopts generative AI to become more efficient may see some cost savings, but others may look to generative AI to create new business models for their business. The latter option could be more successful down the line. 

Kurian told diginomica that relying on your system of record vendor to build out generative AI experiences that create a competitive advantage, is likely a losing strategy. He said: 

Most people choose systems of record for things that they don't see offering a competitive advantage. It's mostly about automating an existing process that they need to be reliable, and they need a source of truth for, but they don't see it as differentiation. Take a practical example, ERP, meaning accounting. 

It’s an important process, you need to count your money. But is a company going to differentiate in the way it competes on the way it counts money? No. And I can tell you that because I've written them myself in my prior life.

It's an important system, but it's not a source of differentiation. So we see a lot of people choosing to use models from providers who are integrating into their own toolset, like accounting, HR, etc. But then using our models, along with others, in places they want differentiation.

For instance, if you look at the work we're doing with Mercedes Benz, they want to see that the products are differentiated in the way they're marketed, the way they're sold and discovered online, the way they are serviced, and the way that people experience it in the vehicle. So they're using our capability for that. 

And so if buyers are looking to differentiate themselves with AI, by adopting an AI platform that allows choice and the ability to build, what does that mean for their backend purchasing decisions? Firstly, Kurian was keen to point out that my assessment of how people interact with systems in the future is still up for debate - conversation may not be the standard. He said: 

First of all, I wouldn't assume that the state of the future is going to be conversational alone, right? Because conversation is very natural when the model responds with text. But imagine the model is generative video, it's hard to talk back to a video. 

I think you will see as models become much more multimodal, in both input and output, the forms of the way that people interact with those computer systems will change. In the same way that you have five senses, right? 

I think, first of all, the interface between people and computer systems is changing. The conversational way started because the first set of models only did text and chat, but we'll see many new things. That's number one. 

But secondly, he does see backend systems becoming further abstracted away from these front-facing interfaces that are facilitated by AI. Kurian added: 

Number two, the way that they change how processes work - we're seeing a lot of organizations modernize their systems. And when I say modernize their systems, they put in what's called an API layer, on top of a set of systems. Abstracted using an AI front end. 

Think of it as a conversational agent, just a simplification. And then over time, once people stop interacting directly with that system, they're able to then re-architect that system gradually in the backend. 

My take

A very interesting week with Google Cloud. I must say, I’ve been impressed by the offering that the vendor has built up and the customer stories I’ve seen in action whilst I’ve been here in Las Vegas. If last year was about AI in theory, it’s been quite interesting to see how quickly some buyers are moving ahead with these generative AI applications to build out new experiences. If this event is anything to go by, the hype is being translated into real-life stories (albeit early ones). It will be interesting to see, over time, where buyers place emphasis. Whether or not they adopt a horizontal AI platform like Google Cloud to compete, or whether they rely on their current systems to integrate AI to accelerate how they operate. And it will probably be a mix of the two. What I think will determine this is to what extent generative AI changes business models. Is generative AI about efficiency, or is it about boosting the top line? I think if it’s the latter, then I think Google Cloud has a compelling proposition. 

A grey colored placeholder image