AI Investment and Strategy: Navigating the Future with CBRE's Chief Digital & Technology Officer

Join CXOTalk episode 822 for a conversation with CBRE's Chief Digital & Technology Officer, at real estate giant CBRE, about AI strategy driving business transformation. Learn how CBRE leverages an "AI Trinity" for efficiency, growth and leadership in real estate operations.

50:39

Feb 02, 2024
2,906 Views

In Episode 822 of CXOTalk, Michael Krigsman hosts a discussion with Satnam Singh of CBRE, the world's largest commercial real estate services company, focusing on their strategic approach to AI investment and integration.

Episode Highlights

Overview

  • CBRE is the world's largest commercial real estate services company, with 115,000 employees globally.
  • Satnam Singh oversees digital and data strategies, applications, and solutions across CBRE's various business segments.

Leveraging AI in Real Estate

  • CBRE applies AI across the real estate lifecycle - from asset identification to investment decisions and property management.
  • They leverage 39 billion data points from 300 different sources to power analytics and AI.

Strategic Pillars for AI Investment

  • Aligning AI initiatives with business priorities and outcomes
  • Achieving faster time-to-value through data, partnerships and capabilities
  • Scaling AI solutions across segments and geographies

Focus on Practical AI Use Cases

  • Emphasis on use cases tied directly to business outcomes rather than AI for its own sake

Cultural and Operational Aspects of AI Adoption

  • Importance of strategic alignment, experimentation culture and responsible innovation
  • Governance for AI recommendations and ensuring data integrity

Overcoming Executive Resistance

  • Understand underlying reasons for resistance
  • Address concerns through strategic problem solving and clear communication of AI benefits

Key Takeaways

  • Strategic pillars for AI investment focus on alignment with business priorities and outcomes, achieving faster time-to-value through data and partnerships, and ability to scale solutions across the organization.
  • Emphasis is placed on practical AI use cases tied directly to business outcomes rather than pursuing AI for its own sake. Data capabilities, AI platforms, and contextualization into industry workflows are critical.
  • Cultural change involving strategic alignment, an experimentation culture and responsible innovation underpins effective AI adoption. Governance, data integrity and partnerships also play key roles.
  • Business strategy guides AI investment strategy with focus on priorities, impact, and ability of AI to address complex challenges and transform workflows. Adopting a pragmatic, business-focused approach is key.

Episode Participants

Satnam Singh is Global Chief Marketing Technology Officer at CBRE and Chief Digital & Technology Officer for CBRE’s Advisory Services business. Before joining CBRE, Satnam served as the Chief Product Officer for the Corporate Tax and Trade division at Thomson Reuters, where he led the development and launch of innovative products and platforms that leveraged cloud, data, and analytics to transform the tax and trade landscape.

Michael Krigsman is an industry analyst and publisher of CXOTalk. For three decades, he has advised enterprise technology companies on market messaging and positioning strategy. He has written over 1,000 blogs on leadership and digital transformation and created almost 1,000 video interviews with the world’s top business leaders on these topics. His work has been referenced in the media over 1,000 times and in over 50 books. He has presented and moderated panels at numerous industry events around the world.

 

Transcript

Michael Krigsman: Welcome to CXOTalk Episode 822. We're exploring strategies around AI investment at CBRE, the largest commercial real estate services company in the world. Our guest is Satnam Singh, the company's chief digital and technology officer.

Tell us about CBRE and tell us about your role.

Satnam Singh: We've got about 115,000 employees and about more than 500 offices in more than 100 countries.

We serve 95% of the Fortune 100. My role, Michael, is twofold. I serve as the Chief Digital and Technology Officer for the advisory services, part of the CBRE, where my team and I, we focus on digital and data strategies, applications and solutions across brokerage sales and financing, property management and the appraisal businesses. I also lead the digital and tech for our marketing organization across all segments of CBRE.

Michael Krigsman: Can you give us context about real estate and real estate services so that we can understand your your strategy within that?

Satnam Singh: At CBRE, we really look at the entire real estate lifecycle, right? So, we look at all the way from asset ID, you know, investment decisions, property management.

We look at operational workflows, you know, how we can introduce AI within those operational workflows to serve our clients better. We look at predictive analytics as I talked about, you know, especially as you look at all the data that we collect, we collect about 39 billion data points across 300 different data sources. So, there is a lot of opportunity, if you will, across the real estate lifecycle where the amount of data that we have, we could really bring a lot of both analytics as well as AI to bear, if you will.

Michael Krigsman: So the data aspect of it is very foundational.

Satnam Singh: Absolutely. We have for example, I'll give you are smart facility management solutions, right? We have about 1 billion square foot of across 20,000 global workspace solution clients that we have. We have building operations and utilization data and highly integrated data. Right. That that we utilize through A.I. and using that you know we started looking at actions such as automated maintenance, if you will.

Similarly, on the unstructured side, we have data through a lot of documents, right? I mean, there's a lot of documents in commercial real estate, right. Whether that be a lease document rental, pitch decks, if you will. So we have, you know, a large foundation of both structured and unstructured data.

Michael Krigsman: How does that data then feed into the underlying strategy that you have for AI?

Satnam Singh: One is how do we really think about AI at CBRE?

So, if you think about the digital roadmap, how do we align that with strategic business priorities? How do we think about the client outcomes? What are our focus areas? What are the market dynamics, if you will, that are important for us? So that's the strategic business aligned. The second is how do we get faster time to value right? What data do we have data that we can use or data that we can source through our partnerships, if you will?

In addition, what capabilities do we have? What capability abilities can we develop as well as acquire again through partnerships or otherwise? And then how do we operationalize faster versus the competition? So that's the second pillar of faster time to value, but it's part of the digital roadmap. The third thing is scale. Michael the size of the problem and the universe talent, uber salary, if you will, of the problem.

Size means, you know, it's easier to solve things of the scale of one property or one lease, if you will, right? But when you really start thinking about building platforms, if you will, for multi segment solutions or operating a portfolio scale of buildings, you've got to think about this differently. And again, at CBRE, we think about the data platform, and this is why we take a very balanced and pragmatic approach, right?

As I talked about the strategic alignment, we really have practical air use cases that we work on with the business teams. Our approach is an eye for the sake of the AI or innovation for the sake of innovation. It's about really utilizing AI and technology in practical situations.

And that's where we look at. What are the opportunities for automation or the operating, you know, operational workflows, if you will? And then as I said, right on the scale side, again, going back to the data platform, having 39 billion data points across 300 sources really is a huge benefit for us.

Michael Krigsman: We’re describing briefly the use cases, so your use cases then are aligned to specific business goals or specific outcomes that are important strategically for CBRE. I'm assuming correct me if I'm wrong, that that's what's going on?

Satnam Singh: In real estate, you have three things, right? You have transactions with a lot of transactions. You have assets which require maintenance and you have assets that are, in essence, investments. Right? So when you think about these transactions, right, they have documents, right?

As I was saying, there are leases. For example, you want to extract data from these leases, such as square footage. The price for square for the rent and so on. Second is, you know, as I talked about maintenance, right, things such as cleaning of property management, if you will. And the third thing is investments where, you know, you think about lots of data coming together, both structured and unstructured, and you have both data at a macro level as well as insights that you gather on a more local level, if you will.

Michael Krigsman: How do you think about AI and in ways that are different than other kinds of technologies?

Satnam Singh: you know, if you step back and you say, look, there are many things that you can do with structured data that necessarily you can build predictive insights and you can get predictive intelligence, if you will, but it really comes in handy.

And we need to step back and say, look, if I as a flavor has been around for a while, right? I mean, there's two aspects to feeling like this is my personal way of really thinking about it, which is that when we talk today about AI, we generally talk about generally the AI or we think about generally the AI, which is about, you know, how do you develop content, new content such as text images or video by giving a variety of input.

Now, previously we've also had the AI, which again, my personal term, if you will, for it is analytical, right? Where you've had for a long time, you've had the structured data that financial companies or companies in retail or transportation have used, right? So for example, if you think in retail personalization and commerce, Right. Is there right. And you know, there are companies look at, hey, how do I think about the placement on messaging generative the AI is about really looking at creation of that documents or analyzing those documents.

And again, when we look at it from a perspective of a commercial real estate firm, right, we look at one of the opportunities, if you will, to create value, to create that productive enhancement by looking at things such as lease documents, if you will. We in our own internal analysis, if you will, our internal work that we've done, for example, looking at manual review of lease documents, we've been able to use AI to cut that manual time by 25%, if you will, using the AI to extract that data.

And similarly, there are other documents such as inspection reports, if you will. And then there is the other opportunity around, you know, smart being smarter about facilities management, if you will. Right? So again, it's the sum of the foundational aspects will remain the same. Michael It's again about contextualize and get right. How do you bring that context, if you will, to the real estate workflows and the real estate operations, which, by the way, is no different than taking a AI and contextualizing it for any other industry.

At the end of the day, you have to take a platform, if you will, a large language model or a predictive capability and make sure you effectively contextualize it in your specific workflows, in your specific industry, if you will.

Michael Krigsman: When you say contextualize that, can you elaborate on what you mean by that and what are the pieces? How do you go about that?

Satnam Singh: A lease document has a different kind of data points than, let's say, some other document we'll have.

Right. And so how do you understand through an LEM model that you want to be able to extract specific items such as the square footage, if you will, or you want to extract something like the price per square footage or you want to extract the address, if you will, of the property that this lease that's about right in there, you have to make sure quote unquote, call it the right prompt, the right entity that you're trying to extract.

How do you contextualize that for real estate? Right. And that's what I mean by is that there is a certain type of call, a dictionary, there is a certain kind of call it entity, if you will, that you are trying to recognize from that document, if you will. Right. And you have to be you have to think about the right set of prompts.

You have to think about the right set of query quoting that document, if you will, to be able to extract information that is relevant to your operations as a real estate company.

Michael Krigsman: Is it mostly around efficiency, the improvements that you're looking for, or are there other use cases as well?

Satnam Singh: it's not just purely about efficiency, right? It's again, when we look at this, we look at it from a perspective of productivity. Enhancements are definitely one of.

That's right. And productivity enhancements means we have to look at the workflows that we have. Right. So for example, Michael 822nd episode of this podcast, right. I'm sure you have a certain workflow that you and I were talking about when before we went live, right? You have a clearly defined workflow and as you look within that workflow,

you know what opportunities exist, if you will, to optimize that workflow for your benefit, right?

So similarly, we look at that from a productivity perspective. In previous lives, I've had the fortune of being in other industries such as tax and insurance and others. Right. And again, you look at the workflows from that productivity and has now actually give you an interesting story. When I was in the music industry a while back at a company called Snow Camp, which was next Napster, second innings, if you will, right.

I found out that, you know, music companies would actually call the the different stores like Tower Records, if not a lot of people maybe aware that there was a company like that. But the analysts sitting at the music label would actually call the Tower Records all across the US and call them and say, how many how many volume of this did you, Bobby CDs, did you sell of this album, if you will?

Well, now you can get that data with digital distribution. That data is much easier to get. So, what do you know, now you can focus on more value generating, more higher level activities, if you will, insights that they had that analysts can do. So that's definitely the productivity announcements. The second thing is you look at the operating model, right within your existing business, what operating models can you improve or can you transform business models, if you will?

Right. And that's where he's really coming at it. A slightly different example, if you will. But in China, I was actually just reading up a few days ago, very interesting. And in China, you have these e-commerce marketplaces that have a lot of influencers, right, selling goods. And what they're doing now is you have these virtual avatars of these influencers that can get ready for about, you know, $1,000 or 1100 dollars.

And they need just one minute of video. And these virtual avatars are up at 3 a.m. in the morning, 4 a.m. in the morning, selling the goods. And if you pay a little bit more, they will actually read the livestream of comments and reply to them. So very interesting, you know, in terms of how AI is introducing that opportunity for new operating models.

And then the third piece is around how do you use AI? And we look at it to say, how do we use A.I.? For my marketing leadership or bringing new thought leadership to the market, you know, things like sustainability or climate tech, if you will, which a lot of you know, our clients are interested in and are baking into their forward-looking plans.

Michael Krigsman: Please subscribe to the CXOTalk newsletter. Subscribe to our YouTube channel, leave comments and check out cxotalk.com. We have incredible shows coming up.

You mentioned earlier the issue of culture and we've been talking about data and the role that data plays and being able to drive these kinds of changes. Where does culture fit into this?

Satnam Singh: You need to think about what I call the digital roadmap and then strategic career.

You need to think about your culture, if you will. Right. And I think within culture, there's a couple of aspects that stand out. First and foremost is definitely, you know, what I was talking about earlier is how do you really think about that? The strategic business alignment you really bring this, isn't it? You know, it isn't just a technology tool.

You know, if we're thinking of anybody who's thinking about AI purely as a technology answer isn't doing justice to the isn't thinking about that the right way and is definitely not approaching it the right way, because then they are probably doing tech for the sake of tech or, you know, innovation for the sake of innovation. If you have a if you think you have a hammer, then everything looks like a nail, as I said.

Right. So that I think is the first and foremost thing is how do you really think about AI as an opportunity to solve something differently? Right? You previously could not solve it because you didn't have the right without it. You didn't have the right infrastructure or the complexity was so large, right, that you can solve. So that's one is how do you think about contextualizing it again within that business, within your strategic alignment?

That's one, I think. The second is creating a culture around experimentation, like with any new technology, right, if you will, And I mean this in the in the context of generative AI, if you will, right? Not everything will be a slam dunk, right? So what you have to do is create a culture of experimentation about learning fast and most importantly, moving at pace and urgency so you can operationalize even faster.

Right? So that's the second pillar, if you will. The third pillar in my mind is what I call responsible innovation. It would be the one thing you have to be mindful about careful about is the potential for hallucinations. Right. If you ask the wrong question, if you will, that is not properly contextualized, then you're going to potentially get some wrong answers in there.

Which means that this is not a silver bullet to everything. Right. You have to think about validation. You have to think about you have to think about validation of process with some human in the loop, if you will. Right. So that's what I mean by culture is right. You have to step back and say, How do I really get started, if you will? How do I really operationalize?

Michael Krigsman: So, you're thinking very explicitly then correct me if I'm wrong about your processes and changing your process needs to reflect the realities of AI.

Like you said, like with large language models, the potential for hallucinations.

Satnam Singh: it's not about doing it for the sake of it, but we're actively saying, look, as we look at these workflows, if you will. Right. What are the possible areas for those API interventions in these workflows?

Right. And we you know, if you look at workflow all the way from your identifying an asset, if you already or some kind of brokerage where there's some leasing discussion going on, you're trying to figure out, Hey, I have certain parameters for which, you know, that are related to this asset that I'm interested in a, you know, office, industrial, whatever it has, retail and so on.

And within that there is a certain amount of workflow, right, in terms of identification of that asset. Then you have, okay, I've identified that asset. What do I think about the financing of that asset? Or if you are a seller, how do I think about the sale of that asset? Right? If you are acquiring an asset, then you know there is an aspect of managing that asset, both managing as in physically managing the asset, but managing the investment that you've made into the asset.

And that's what I was talking about earlier. If you recall, the three things that I said was, okay, you know, asset identification and there's documents around it. And as you think about those documents, you know, what do you think about how do you think about the workflows where you can create some efficiencies or productivity enhancements, if you will, as you think about property management, You know, how do you think about, you know, enhancements there?

And then as you think about it as an investment, what are the opportunities? Let me take an example, if you will. So again, going back to our Smart facility management solution, right, which I talked about, has about we have about 1 billion square foot using the building and operations data. Right. This is part of that contextualizing, if you will, the workflow, building it using the building operations data.

We looked at actions such as automated, automated maintenance, right? And so that meant that we could reduce the tech dispatch, if you will, by about 25%. And the operating expenses that were there for this automated maintenance, we were able to reduce the energy and the maintenance by as much as 20%. So again, we look back and say, let's look at all this workflows, if you will, and look at opportunities where you could bring II into the mix and that can actually create value for it for the organization and create value for our clients and deliver much better differentiated outcomes for our clients.

Michael Krigsman: So on that point, we have an interesting question from Lisbeth Shaw on Twitter who asks, Can you provide some examples of how AI has changed the way real estate services, the real estate services and investment business works?

Satnam Singh: as you look at some of these interventions, Right, Because I mean, it's a great question, but again, it spans across multiple aspects, if you will. So, if I go back to what I said right. Is you look at things such as documents.

And so, for example, we have today internally created automated content creation capabilities, right. That take us that used to take two weeks, but now take 2 hours. You know, I talked about, you know, again earlier about taking this smart facilities and and, you know, saving operating expenses such as energy and maintenance by as much as, you know, 20%, if you will.

But again, similarly, we had a health care client where we utilized dynamic cleaning and we saved 11% savings over the baseline that we have. So, again, it's the opportunity to really look at that data, look at that workflow and say, where exactly do you bring into the middle here, into as an intervention, as an opportunity, if you will, to really create some workflow productivity enhancements and, and some savings, if you will?

Michael Krigsman: Can you drill in a little into the mechanism? What is it about these tools or the technology or the or the way you're using the data that, for example, enabled you to go from what did you say, two months to two weeks or something like that?

I don't remember the exact numbers you said.

Satnam Singh: About two weeks to 2 hours, right? So in a sense, what you are doing is you're looking at the workflow and you're looking at the existing set of documents that you're looking in, in essence some form of pattern, if you will, to say, look, the more I can learn about this and to a certain extent extend.

Michael, it's also about the scale. How do you really think about it? So when you have a scale, then you can identify certain patterns in those patterns, start giving you insights into, you know, what is that workflow, repeatable workflow that is happening here and how you can how you can create that workflow or operationalize or optimize that workflow.

And in essence, you have to have three things from a foundational perspective that you have to bring into the picture here. One is the data capability. So, you know, it doesn't happen overnight, right at CBRE, our focus on data and having enterprise grade technology has been there for a long time. Right. Which has allowed us to move quickly, you know, if you will, into AI, especially junior area like.

So that data capabilities having an enterprise data platform where we can transform data at scale, understand it at scale, get insights from this data scale as a key pillar of our foundation, if you will. Right decks on. The second thing is really looking at is setting up a generally the AI platform and, you know, not necessarily trying to solve a one-use case if you will, but understanding, if you will, patterns.

I mean, we've launched industry's first large language model, if you will, multi-language model interface, right? And we've got we focus on rather than focusing on on one specific use case, we've made it self-service internally so that people could utilize it. It has many of the same features as CBT. We made it in such a way that people you could utilize it for many different use cases, if you will, right?

And then the third piece is how do you really go back and look at utilizing the information, the insights that you got from those deep data capabilities to say in this workflow, what is the end result that needs to be there, right? And what does the end result really mean for the customer that is utilizing that end result?

So I think across those three things, the data capabilities, some form of analytical A.I. to really think about the usage and then the Jenny AI platform, if you will, you start looking at that pattern recognition, you start looking at that workflow, and it's not a one size fits all. And, you know, situations are different, but, you know, once you have a certain approach to a certain pattern to it, it starts with giving you that value, if you will.

Michael Krigsman: We have a really interesting question from Arsalan Khan on Twitter. And Arsalan is a regular listener and he always asks very, very thoughtful questions. You're working with such a large body of data. How do you ensure that you have confidence in the data? Because you're ultimately going to be making potentially very large decisions on the basis of that data.

Satnam Singh: We have, you know, a lot of internal data that we would triangulate with where we have data Partnerships are real, if you will. Right. And those data partnerships, what data we have, what data can we use, and third party data, if you will, data that we gather through partnerships allows us the ability to be able to triangulate some of that data and understand the call it the integrity of the data.

And if you go back, if you will, right where I said was look, as in when you set up a a culture, not every use case will be successful in the beginning. Right. And in some cases, we've actually had to we've learned fast that there are certain type of data, if you will, that either isn't collecting, collected at the level of granularity that is required for this use case or isn't being collected at the level of comprehensiveness, if you will, in in this.

And then we step back and said, okay, what are the you know, what are the proxies, if you will, for that kind of dataset that we can use instead of, you know, the scope that we were originally thinking about? Could we think about it in a smaller scope, if you will? Right. And so that's where we've stepped back and said, most important in our focus has been, you know, our focus previously and continues to be around as one key aspect of our data platform.

This is again, I stressed back on the enterprise data platform that we have is we have to step back and say, can we enable this unique use case through the data that we have access to? Are there opportunities for proxies? Are there opportunities that we need to fundamentally think about the workflow, not just in terms of the workflow, the existing workflow, but do we need to change the workflow that we have in order to be able to collect this data going forward?

Or are there proxies, if you will, that can give us an opportunity to make some advances in the use case that we have? So but a great question. You know, if that's a very key, I would say path forward in focus, if you will, that we have in terms of how do we really think about our data, both what we have, what we can use and that which we can source, if you will, to our partnerships?

Michael Krigsman: So, the partnerships aspect is also another very important foundation of your ability to execute on this kind of a AI data driven strategy.

Satnam Singh: Absolutely. Absolutely. Michael. I mean, we don't believe that, you know, we just have to go it alone. Right? It's it's we have a very strong build by partner strategy, right? I mean, we don't believe that. Again, we have to do it. All right. At the end of the day, it's about the time to value and the client outcomes that we really focus on.

And to that extent, you know, again, going back to that strategic alignment that I talked about is we look at those strategic CBRE priorities and we say in order to achieve these, in order to get to these strategic priorities, what do we have as capability today? And who are the potential partners that we need to have in the marketplace?

And we scan the market on a regular basis for partners as well as emerging companies, if you will. Alison Bell, who I work very closely with, you know, is a global head of our digital strategy, tech acceleration and partnerships. And Alison and her team, you know, regularly look at the market from, you know, from our digital strategy perspective for what we need in terms of partnerships.

And today I will say that, you know, we have partnerships with all the major tech providers, if you will, and it has given us an edge, you will, in terms of early access to capabilities such as large language models or other gen AI capabilities that we have integrated into our ecosystem of the digital and technology stack that we have.

I think what we've also made to share, we've also made selective investments, if you will. So we made a $100 million investment and invites. We were in the leasing and property space as well as, you know, partnership with Key, if you will. Deep is a company that collects energy, waste and water data. So if you think from a sustainability perspective, that's very important.

So again, to bring it back to your point, again, it wasn't it isn't about, you know, building everything internally, if you will. It's about the time to value to the market.

Michael Krigsman: The DNA of CBRE, of course, is about commercial real estate, and you're describing ways of working and ways of thinking about this data, this data asset that is very different from the skill set and the background and the culture of commercial real estate.

So how do you make that kind of transition where the folks in the company can learn to think this way and learn to think about data as being an asset?

Satnam Singh: Yes, it's relatively new technology. But if you think about other previous technologies, Right. I mean, digital transformation, right. About utilizing data at scale, if you will.

Or about being able to capture other data in a large scale data at scale, right. I mean, at CBRE, you know, we're all about physical assets,

And physical assets have a lot of data associated with them. Right. The location, you know, location being the address, if you will, or the floor or within the floor of what exactly, you know, if you have more than one, you know, company on a given floor, you know, what exactly is part of one versus the other or previously even Iot.

So, I think it ultimately goes back, if you will, to say, hey, what exactly are those strategic priorities? And talking about, you know, how you think about AI as a potential solution. So again, we need to step back and this is something that we have focused very much on, is stepping back and really defining the problems statement.

And so once we define that problem statement, that's when we bring it into the mix and say, is the AI the right solution here? Is is a different technology the right solution here? Right. We don't have to, in essence, bring it to bear on each and every question, if you will. Right. In some cases, for example. Right.

I mean, if we're trying to build a simple dashboard that gives people, you know, certain metrics. Right. Simple, quick metrics. Right. And the first, you know, you just want to be able to quickly build a dashboard and then you say, okay, how do I make it better, Right? How do I make interaction with this dashboard better? That's where you utilize a chat bot like capability to be able to, you know, query the data in the dashboard.

Again, it's, you know, I would say there's a bit of maybe a bit of a misunderstanding out there that, you know, commercial real estate, you, you know, isn't as far advanced, if you will. I think it's all about, you know, our applicability, if you will, of digital transformation or AI, if you will, to the solution. The one other thing you kind of touch based on, and it's a slightly bit of a digression, is, you know, there is this aspect that every company needs to be a tech company and I have a bit of a fundamental call it a personal disagreement with that.

Which is to say it, not every team needs to be called a tech company, right? Not every company needs to be called a tech company. Right. It's about how do you really utilize tech in order to operate within the workflows, in order to operate within the industry that you are working on? Right. We are a commercial real estate company.

We bring technology to bear. We bring it to bear to really help our clients achieve certain outcomes for us to achieve certain strategic business priorities. Right? And that's one of the many tools that we have at our disposal to be able to achieve the outcomes that we need to achieve.

Michael Krigsman: think you're being very clear that the fundamental strategic goals of the company have not changed because a guy has been introduced into the mix. And so everything you're doing with these tools continue to support the ongoing strategic goals of the enterprise.

Satnam Singh: Exactly of our customers. At the end of the day, it's about, you know, I know I've said strategic business alignment.

Other part of that, I said client outcomes. At the end of the day, it's about how do we serve our customers better that customer centricity, right? And if you do a great job at being customer centric, if you do a great job at really focusing on solving the problems that your customers are really facing, thinking about or need to plan for, then I think you set yourself up in the right for being successful.

Michael Krigsman: Arsalan Khan comes back and says, How do you know? Or figure out that the recommendations made by the AI based on the data are correct or not? And he's asking if you have any kind of a governance process in place for this, especially given the fact that you're working with so many different partners and you're sourcing data from different organizations.

Satnam Singh: One is about data and the other is about validation of the outcomes rate of the recommendations, right? So one is about data and the other is about, you know, the model that you appointed to that data. You know, is that model really giving you the answers or is serving you appropriately?

In terms of the interest rate? I think from a data perspective, as I said, right, we have the benefit of having a large treasure trove of data, if you will, of 39 billion points. So where we are, you know, as the largest commercial real estate company and a bit of a unique situation, if you will, to to triangulate, to validate some of that data and be able to utilize both internal data that we can use as well as the external data partnerships.

So, there's that validation, if you will, from that perspective. I think the second part is about model and models are iterative, right? So you look at and say, look, that model recommendation that I got right, you, you iteratively build a model, if you will, from a small scale to operating at a larger scale. And this is what I was saying earlier, right?

It's easy to operate at a small scale, but it's complex and to operate at a large scale. And so from that model perspective, you really, really look at iterative approach, if you will. So, you say, look, you know, let me apply this particular model in one particular building, if you will, with one particular client, if you will, and see what's results.

Do I start to get right Because inevitably what you will find is and this, by the way, is not specific to commercial real estate. This is very broadly What you do is when you're building a model, you will find that there are certain aspects, certain factors that you did not take into account or you did not know to take into account.

And that comes when you're trying to operationalize the model, right? Because it's one thing to build a model and then it's another thing to actually put the model in practice and operationalize it. And so from that perspective, we, you know, we have constant feedback loops, constant validation loops. And this is where I was saying earlier also, especially in the context of generative the AI, you have to be careful and you have to have that validation and you have to have that human in the loop, if you will, to make sure that you know, you're not getting in some form of hallucination.

I know I'm digressing a little bit from a model, if you will, to do generative AI, if you will. Right. But again, this is about creating the right feedback loop, making sure you understand the right business problem that you were trying to solve with the model, making sure that you have the right data points in order to assess the right business problems and to assess that the model is actually performing on.

I'll give you an example, which I can't talk much in detail about, but at a high level, you know, we were working on we've been working on this particular model, if you will, for one of our businesses. And the question one of the feedback that we recently got was, yeah, you know, we can look at documents from this resolution or this perspective.

What if you change the document orientation, What happens? And again, this is exactly we already in that operationalizing the model really comes into place, right? So hopefully I've asked Arsalan, I've answered Arsalan’s question. You know, I think Arsalan, you should definitely connect with me on that because I'd love to learn more. And I think anybody who has a great thought process, a very clear, structured thought process, I'd love to connect with.

Michael Krigsman: We have another question from Twitter, and this is you've been talking about the relationship between the company's business strategy and your A.I. strategy, but how do they influence each other, your business strategy and the AI investment strategy? How do they work together?

Satnam Singh: You're not out there to say today I'm going to have a use case right? I mean, because if you go out there and see IVR use case, then you are simply going to look for a place where you can go invest it.

But you need to step back and say, well, what is the business strategy? Where can I go solve that business strategy better, faster, in a larger scale? How can I create that impact, you know, for faster time to value? How can I be more cost effective? I drive higher productivity or can I bring new solutions, if you will, to bear because they were previously very complex, if you will.

And so in from that perspective, you say in order to really achieve that business strategy, diverse set of tech investments and as part of the tech investments, these are the set of investments I need to make to be able to achieve those three things and faster time to value ability to deal with larger scale and be able to solve more complex problems.

You know, there is this blog post and that this reminds me about Bill Gates has those Gates notes that subscribe to and they land in my email. You know, every whenever she publishes in one blog post from about a year ago that he wrote and he said, you call it the age of A.I. has begun. Very interesting. It's it's a very comprehensive one.

I would suggest people to look, you will find that age of AI has begun and read it is this you know, he talks about among other things, he talks about the fact that, you know, the use of AI is especially relevant in systems that have scale and complexity that is tough and hard for humans to really deal with.

And he talks about the example of he talked about the example of drug discovery. Right. Or health care. Right. In which you have complex biological systems. Right. That are very tough for us as humans to be able to do, be able to track and comprehend on that level. Right. So, I think, again, I go back to the scale and complexity.

And in that context, second thing is personalization is what he talked about, right, especially within the context of the societal good about education. Right? How do we really tailor content? How do we how do we track, you know, how much engagement a student has with a certain content and how do we track it so we can tailor content for better educational outcomes, if you will.

Michael Krigsman: How do you think about the ROI of your investments in AI? Is there anything are there any distinctions how you evaluate air ROI relative to other investments that the company might make?

Satnam Singh: There's one aspect, culture of experimentation, right? Learning fast sometimes, you know, that precision that that you can have with certain business cases to be able to say, look, if I were to go down this path, this would be precisely the impact I could make and this would be precisely the outcome that I could get and therefore precisely that return I could get right with.

As with any new technology and with AI, you may not be able to get that level of precise answers about what the impact of that scope will be. And so I think fundamentally, no, there is no difference, right? When you think about it, the fundamentals and what I mean by the fundamentals, what you are tracking to make sure that you look at the ROI, which is things like, you know, productivity or, you know, top line growth or things like free cash flow that can be generated or client outcomes like better client experience, if you will.

Fundamentally, those metrics should not change. Right? And we don't think about it. You know, they must be different.

But the aspect is that if you look at 180 use cases, you may be able to bring that precision to a certain number of them in and in other cases, you just have to go forward and say, look, you know, this is about experimentation, this is about learning faster and this is, you know, if we don't go through this door, we're not going to be able to understand what lies on the other side.

And so hopefully that answers your question. Michael. It's yes, fundamentally the same. But keeping in mind that when faced with a new technology that you have to make, you have to take certain you may not have the most precise answering part of the use case and if you will.

Michael Krigsman: So, part of that is the cultural change responding to the nature of the technology.

Satnam Singh: Yes. Exactly. Exactly.

My advice for anybody would be to say, look, you know, as you look at the opportunity for it to drive impact, especially transforming, you know, the way in which we work or in terms of helping us gathering insights, driving predictions, we just have to step back and say, no, we need to we need to have a culture of learning fast. We need to have the culture of moving and pace and urgency. And sometimes that may mean that you seed a separate investment for these kind of things so that you can do that. And equally importantly, something that we've done as CBRE for a long time is, you know, making sure that you invest in the right data in enterprise grade technology stack and then finally is making sure that you continue the context, realize your problems and your solutions.

You contextualize your solutions to the use cases, to the industry that you are in and the workflows that you.

Michael Krigsman: Being clear about the outcomes, the processes and the workflows, as you said.

Satnam Singh: Exactly, exactly. And you know what? It sounds easy, but it's not like because sometimes, you know, there's this, hey, let's go shade change the shiny object, if you will. And that's what I'm saying is let's not chase the shiny object.

Let's figure out where that application is and what is the value of that application. Let's continue to move forward, but be mindful, be pragmatic, be strategic about it.

Michael Krigsman: Actually, it does not sound so easy to me because you have an existing set of clients, set of processes and revenue drivers, and the company works in a particular way. Now you're talking about introducing a potentially significant change which is disruptive, but at the same time you don't want to disrupt what works.

Satnam Singh: Yes. And that goes back again to what I was saying about operationalizing. Right. I right. You have to think about in the context of how do you really bring the the technology you bring, if you will, bear in that operational context And what are some of the aspects you need to be mindful about, you know, when you try to operationalize it?

So, it is this is again, why it goes back into you're looking at what are your strategic business priorities and what are the workflows, if you will, right, that you have? Right. I mean, you know, in my prior lives, I've served as a chief product officer and I, I, I often talked about that. You know, I want to take the work product management and change it to workflow management because at the end of the day, you're building products for a certain workflow.

You're trying to solve for a certain improving sort of workflow. So should we call it workflow management with digital technology and or is it digital and technical digital applications and write the best technical technology enterprise grade technology Stack Yeah, maybe part of that is that right?

Michael Krigsman: We have one final question from Twitter again, from Arsalan Khan, who again asks a very thought-provoking question. And it looks like he's going to have the final word here or the final question. What happens when an executive uses their vote to not do anything with AI inside the organization? How can non-executives change that veto?

Satnam Singh: You have to step back. It's first and foremost understand why they did that, right?

Because if you're trying to solve simply as a knee jerk reaction and say or look, you know, somebody said such and such thing and therefore I need to have an immediate reaction because, you know, my belief is that the problem requires a different solution. Then you're going to find yourself in a tough spot. I think the first and foremost question, the first and foremost approach is to understand why that veto happened in the first place.

You know, there might be some legitimate reasons, right? You know, again, as a you know, I always coach and advise, you know, people is to say, first and foremost, understand the industry, understand the workflows, get into the details of the workflows, if you will. Right. And so is it the fact of, hey, that veto happened because, you know, the solution that is being designed is is going to is going to be great, but it doesn't have the right level of operational considerations to it for deployment.

And if that's the case, then you go solve that problem. If the problem is, hey, yeah, the solution is great, but it only solves part of my problem. And in order to, you know, you're asking me to fundamentally change 100% of my workflow while you're really going to bring a solution that only solves 10% or 20%. And I'm not going to go disrupt people who, you know, are used to a certain workflow only to be disrupted for 10 to 20% of the workflow, then that's it.

Then you have to go answer that question differently. So, I'd say the first and foremost thing is the art of asking questions, right? Step back, ask the question, understand the reason behind the veto, right, if you will, because nobody ever said that, you know, hey, I don't want to do things. I don't want to grow my revenue, or I don't want to have a higher margin or I don't want to have higher free cash flow.

There's nobody out there who says that, right? So, there must be a reason behind it. And I think it's imperative, first and foremost, to understand that reason.

Michael Krigsman: Lot of wise advice right there.

And with that, I want to say a huge thank you to Satnam Singh from CBRE. Satnam, thank you so much for being here with us today.

Satnam Singh: Thank you for having me, Michael. It has been a pleasure.

Michael Krigsman: Everybody watching, thank you for being such great audience and to the people asking these excellent questions. Thank you so much. Before you go, Please subscribe to the CXOTalk newsletter. Subscribe to our YouTube channel, leave comments and check out cxotalk.com. We have incredible shows coming up. Thank you so much, everybody. I hope you have a great day and we will see you again next time.

Published Date: Feb 02, 2024

Author: Michael Krigsman

Episode ID: 822