AI Projects: The Big Benefits for Organizations

STEVE HENDERSHOT

Artificial intelligence has been in the spotlight for years, and thanks to greater access to tech tools such as generative AI, it has gone from an out-of-sight advancement to an everyday asset.

McKinsey’s The State of AI in 2023 survey found that 55% of respondents say their organizations have adopted AI, but less than one-third say their enterprises use it in more than one business function. That leaves a lot of room to grow. 

Project professionals have been hard at work, delivering a recent wave of AI successes. And now they’re ready to share the knowledge they’ve acquired to equip a new cohort of project leaders looking to implement AI. So today’s show is like the organic version of a self-improving algorithm. Stand by for the update patch; it’s installing now …

In today’s fast-paced and complex business landscape, project professionals lead the way, delivering value while tackling critical challenges and embracing innovative ways of working. On Projectified^®, we bring you insights from the project management community to help you thrive in this evolving world of work through real-world stories and strategies, inspiring you to advance your career and make a positive impact.

This is Projectified. I’m Steve Hendershot.

There’s nothing artificial about the impact that AI is making right now—across industries and around the world. Project teams are stretching to find innovative ways to deploy the technology, create value and mitigate risks, even as a lot of AI applications are brand-new, and the technology continues to rapidly evolve.

In 2022, the global AI market was worth a little less than half a billion U.S. dollars, according to Precedence Research. Now it’s surging forward on the coattails of breakthrough applications and products such as ChatGPT, and Precedence estimates that the value of the AI market will surpass 1 trillion U.S. dollars by 2027.  

Today we’re speaking with project professionals tasked with harnessing the power of AI while also delivering projects whose AI elements are futureproofed and derisked. Let’s start in Dubai. I spoke with Mohammed Nabtiti, project management office and governance manager at Beeah Group, a holding company focused on sustainability, about the organization’s new headquarters. It’s said to be the first building in the Middle East that features fully integrated AI, from adaptive regulation of the mechanical systems to an AI concierge to smart meeting rooms that recognize visitors and log them in to the integrated presentation systems the moment they walk in.

MUSICAL TRANSITION

STEVE HENDERSHOT 

Thanks for talking with me today, Mohammed. I’m excited about this conversation, so let’s dive right in and talk about your company’s headquarters. This “Office of the Future” has a lot of AI elements, so what was the inspiration that kicked this all off? 

MOHAMMED NABTITI 

The Office of the Future project at Beeah was inspired by a vision that aimed to revolutionize the concept of workplaces. We wanted to set new standards for sustainability and intelligence. For that, the project had two primary driving forces. The first is creating a sustainable workplace. This entailed utilizing AI to optimize building resources and reduce environmental impact. 

Smart IoT [Internet of Things] sensors were integrated into the building’s infrastructure to monitor and manage energy consumption in real time through the BMS, which is the building management system. Then the AI-driven building automation system analyzes data from the BMS to automatically adjust lighting and temperature settings for maximum energy efficiency. 

There is also the digital twin, a live 3D representation of the building that understands and comprehends the spaces in the building and detects in real time the occupancy levels of these spaces, which aids in making energy decisions. Also, the digital twin displays every single asset in the building and monitors their health.

STEVE HENDERSHOT 

So sustainability was that first big driver. What about the second? 

MOHAMMED NABTITI 

The second driving force is showcasing AI and smart office technologies. The Office of the Future aimed to serve as a living example of how AI can seamlessly enhance user experience by making it a visible and an interactive element of the workplace. For example, the AI virtual concierge is a standout feature that welcomes visitors upon their arrival to the building and provides them with information in an intuitive and interactive manner. It recognizes the visitors using facial recognition. It also notifies their host of their arrival. It orders coffee for them and so on. This is only the first step upon the visitor’s arrival to the building, and we have designed a comprehensive arrival-to-departure journey that building visitors go through, which demonstrates the potential of AI in enhancing customer service and engagement.

STEVE HENDERSHOT 

Tell me more about these AI features. Some are running in the background, like the building management system, but others are more visible to visitors and employees, like the concierge. How did you decide which elements to include? 

MOHAMMED NABTITI 

The selection of AI elements was guided by a combination of factors, including the potential for positive impact and alignment with our sustainability goals. We undertook several key steps in building our business case. First, we started in 2018 by benchmarking existing smart and green buildings in the world, including the Edge building in Amsterdam and the Googleplex building in California. We studied the elements that made them smart and sustainable, and [we] created a comparison to use as a benchmark for us to achieve beyond it. Our business case included detailed projections and KPIs [key performance indicators] on how AI could reduce energy consumption, optimize resource management and enhance the overall workplace experience.

For example, AI-powered smart meeting rooms were chosen because they enhanced collaboration and communication among employees. When a visitor or an employee or any user of the meeting room enters the room, the room recognizes this user immediately through facial recognition, and it greets them and welcomes them, just like the virtual concierge has greeted them when they arrive to the building at the entrance. And then the room will ask the user, “Well, should I start the meeting now? Should I record the meeting?” It will start and initiate the meeting on its own. And while the meeting takes place, the AI will automatically record the meeting. It will also convert voice into text. It will write minutes, and it will send those minutes to the attendees of the meeting. We wanted the Office of the Future to become a living example of how AI can be harnessed to create a workplace that is not only intelligent but also environmentally responsible and user-centric.

STEVE HENDERSHOT

What were the conversations like around buy-in? I assume that at a very high level, executive leadership was all in on making this building as high-tech as possible. 

MOHAMMED NABTITI 

We were fortunate to have strong support from our executive leaders and stakeholders, and that has happened from the project inception. However, building buy-in for such a transformative project required clear communication of the benefits. It also required clear alignment with our organization’s values, and we had to have a well-defined business case. We engaged in extensive dialogues with key decision-makers to ensure they understood the project’s strategic importance and long-term value. We engaged with key stakeholders at an early stage of the project’s conceptualization, and this is a key step: Keep your stakeholders engaged from the very beginning and throughout the entire project. This early involvement allowed stakeholders to provide input and insights, fostering a sense of ownership and collaboration.

STEVE HENDERSHOT 

Let’s move to risk management and mitigation. I’m interested in the process because the technology is evolving at a faster pace than the multiyear development cycle for the actual build. How did you approach risk management, and in addition, what were the risks that you identified, and how did you mitigate them?

MOHAMMED NABTITI

As you rightly said, technology changes very fast. It was definitely a high-risk project because it is innovative, which means it is, by definition, something that wasn’t done before, or at least it wasn’t done in this context before, which creates a lot of uncertainty that you won’t find a straightforward way to deal with in OPAs [organizational process assets] or in knowledge bases.

One of the primary risks was the integration of various AI systems and ensuring their seamless operation. Each of the AI systems provides a specific function. Each of these functions works perfectly on its own, but we don’t want them to work in silos. We rather want them to complement each other, and we want data to flow between these systems. For example, the facility management system should communicate effectively with the building automation system. We want the smart parking system to work with the visitor management system and with the virtual concierge, and also work together with the meeting room system, and so on.

So the scope was to integrate between all these independent systems to create the comprehensive and seamless arrival-to-departure journey that we talked about, which created the project’s biggest risk, which is integration. To mitigate the risk of siloed AI systems, we adopted a strategy of creating a system of systems architecture. This involved designing a cohesive layer that would allow all AI elements to work together seamlessly.

So this was one way to mitigate the risk. Another way to mitigate this risk is by simply using agile. Just use agile. Every problem, no matter how big or how much risk it can introduce, can be solved when broken down into bits and pieces. And this is exactly what the agile mindset is all about. Always deliver small pieces in multiple cycles. Always get your stakeholders involved at all times. Always get feedback from your stakeholders and keep gradually building your solution. And, trust me, your risk will vanish.

STEVE HENDERSHOT

I’m also interested in how the team balanced risk and innovation. Because at some point, the building is done, but that’s not necessarily the case for AI. There will be advances to consider implementing in the future. So how did you balance risk and innovation? And what’s the plan for remaining open going forward? 

MOHAMMED NABTITI

Well, there are always continuous enhancements, and this is also part of using agile. We always get feedback, whether from users or whether from the AI system itself. As the AI works and operates in the building, it makes decisions. We monitor these decisions, and we can always change them, fix them, and input new data to teach the algorithm how the decisions should be modified and how to optimize the building further.

STEVE HENDERSHOT 

When using AI tools or deploying AI solutions, there’s that question of who is responsible for the outputs or the decisions an AI tool makes. How did you arrive at that hierarchy?

MOHAMMED NABTITI 

Accountability is tricky, but ensuring accountability for AI outputs and decisions is a critical aspect of the AI project management. Ultimately, the primary responsibility for the outputs and decisions made by AI tools rests with the organization itself. It is the organization’s duty to ensure that AI systems are designed and implemented in a way that aligns with its objectives and ethical standards. The organization should establish a clear governance structure to define roles and responsibilities for individuals and teams involved in the AI deployment, which means there should be a cross-functional collaboration between data scientists, business units, the architects, legal teams and ethical experts.

STEVE HENDERSHOT 

Did you need any new skills or knowledge? What was required of you to take on this project?

MOHAMMED NABTITI 

Yeah. Of course we needed specific skills. Managing an AI project required a combination of technical knowledge and project management skills. It was essential to develop [a] deep understanding of AI technologies and machine learning. This involved learning about the fundamentals of AI, machine learning algorithms, natural language processing, computer vision, and data analytics and so much more. In the context of the Office of the Future, it was crucial to comprehend how the AI within the building learns and adapts over time through its interactions with the environment to improve its decision-making.

It was also important to involve the real experts. For example, JCI [Johnson Controls], which is one of our main partners in the project, have involved stakeholders with PhDs in [the] system of systems integration and data science. Microsoft, on the other hand, has also deployed global experts in smart buildings and experts in the user interface and user experience fields to design how the AI interacts with the users. Working and collaborating with these experts was crucial to bridge the gap between project management and technical implementation. As the project manager, it was important to understand what input data the AI system required to learn and improve their performance. Knowing how often to provide these inputs and setting a timeline for the expected improvements were critical for managing the project’s AI components.

STEVE HENDERSHOT 

The headquarters building has been in operation since March 2022. What has been the outcome so far? How has the organization benefited? And then how have you measured the success of the project overall?

MOHAMMED NABTITI 

The Office of the Future project has yielded multiple positive outcomes. These outcomes were measured through KPIs that align with Beeah’s goals and even with SDGs, which are the [Sustainable] Development Goals. In addition to the fact that the building runs on clean energy through a photovoltaic station within its premises, we were able to optimize resource usage and reduce energy consumption using AI-driven systems. The project became a valuable source of knowledge and experience, and that’s for Beeah and its partners. Lessons learned during the project have enriched the organization’s expertise in AI implementation and sustainable building design and how to go around managing complex projects in the future.

STEVE HENDERSHOT

Tell me more about those lessons learned. How are they helping you in future AI initiatives? 

MOHAMMED NABTITI 

One of the most critical lessons I learned is the significance of comprehensive scoping and planning. AI projects are complex, and when you start identifying your use cases, it is easy to get dragged in[to] the endless possibilities of what AI can do. This is why a well-defined scope is essential to set realistic expectations and ensure successful implementation. Comprehensive planning involves setting clear objectives, defining the scope, building a clear system architecture, implementing the architecture you designed, allocating resources effectively and establishing a realistic timeline.

A second lesson is that AI implementation often necessitates significant changes. These changes can happen in processes and procedures and even in the organizational culture. The lesson learned here is that change management is as crucial as technical implementation. You must prioritize change management as a core component of the project strategy. This includes proactively engaging stakeholders, communicating changes effectively with them, and providing the necessary training and support to ensure a smooth transition. And the list goes on and on, of course, if you want to ensure a successful change management. Basically, you should approach future AI initiatives with an agile mindset of continuous engagement and continuous improvement.

MUSICAL TRANSITION

STEVE HENDERSHOT 

Beeah’s Office of the Future project was all about effective AI implementation in one very specific context. Other AI projects, meanwhile, are focused on ensuring that the tech is set up to deliver value to a broad range of stakeholders and use cases. That’s the case at technology giant Siemens AG, where Sanjukta Ghosh is a data and AI leader at Siemens’ Munich offices. Sanjukta spoke to Projectified’s Hannah LaBelle about how project professionals can master the unique contours of AI projects and the challenges they present. 

MUSICAL TRANSITION

HANNAH LABELLE 

Sanjukta, thanks for speaking with me today. Let’s start here: What sparked your interest in AI? 

SANJUKTA GHOSH 

I’ve always been working on designing and developing intelligent systems— for example, different kinds of sensor-based systems like thermal imaging systems, inspection systems, where essentially the environment is perceived by sensors and the signals are analyzed. And the goal and motivation for me has always been to solve customer problems.

Now, two decades ago, it was about using more classic algorithms, and over the years as AI evolved and showed promise, I started exploring the use of AI as a means to solve some of these problems in the industrial context, and I was amazed by its potential. I remember quite some years back, I was designing a person-detection system, and there was one case where I thought the model I trained was detecting a person falsely because, at first glance, it seemed there was no person in the image. But when I looked closely, I realized there actually was a person that was partially hidden by a tree. And that was a turning point where I got really interested in AI and thought, “Okay, this now has promise, and this is something I must pursue.” 

HANNAH LABELLE 

So what knowledge or skills would you say project professionals need to manage AI projects? 

SANJUKTA GHOSH 

I had the advantage of also being a designer and developer of AI systems. I understood quite well the characteristics and the nature of the work, and this definitely helped me shape and manage AI projects. Having said that, in order to manage AI projects, it’s not really mandatory that one needs to understand the detailed workings of such systems. One definitely needs to understand the nuances of AI systems, like its iterative nature, the different stakeholders involved—there’s really an ecosystem of stakeholders here. The standards and regulatory scene, the facts that along with code, now there’s data and models to be handled, the deployment aspects, and so on, right? Now it definitely helps to understand the technical aspects. So this is something that continues to evolve quite fast in AI. Other than this, the fundamentals of project management still hold, though, and they need to be applied in the context of AI projects.

HANNAH LABELLE 

Next, I want to talk about your work in AI operationalization. You’re part of a team that’s standardizing the execution of AI projects across Siemens’ portfolio. What prompted this effort, and how does it fit into the overall company strategy?

SANJUKTA GHOSH 

In the industrial world, customers are interested in solutions to their business problem, and the power of data and AI is transformative. It is one of the levers for enabling digital transformation and a means to enable solutions to these business problems at a level that was not possible before. What I mean is, if earlier you could analyze, let’s say, data from a single asset like a pump, now one can analyze across many assets, across multiple sites, over a period of time. So in some sense we’re really leveling up in the value chain. And there’s a much more frequent use of AI. 

So for a company as large as Siemens and with different businesses, while on the one hand there is a need to standardize in order to scale and be efficient, we need to be able to adapt to the needs of each of these different businesses. So we also believe in aggregating these best practices, learnings from the various experiences with customers, and not just that it’s a research topic. And as far as, let’s say, the company’s strategy goes, the overall Siemens strategy is to combine the real and digital worlds in the industrial context. It’s important that whatever we design in industrial AI should be interoperable, open and flexible. This means that it’s possible to access and integrate into other systems and operations.

HANNAH LABELLE 

What would you say are the top challenges you’re facing in this process, and how are you and the team working to overcome them?

SANJUKTA GHOSH 

To start with, it really involves understanding the value chain. At the end of the day, any algorithm that we might put in is always going to bring in some inaccuracies, so it’s also about understanding how AI could best be used to bring value. Now having understood that, it’s about having the means, mechanisms, like the tools and infrastructure, to handle really large amounts of data and models along with the code versioning of this automation of the pipelines, handling the traceability of these artifacts over the entire AI system life cycle in a secure manner and economically viable for the business cases as well, right? There’s also things like the availability and accessibility of the data, because quite often connectivity are challenges or interoperability are challenges. You have legacy systems. How do you get the data out? There’s also the inherent experimental and iterative nature of AI. This can be quite challenging, actually, and that’s different as compared to other kinds of projects. And also acknowledging and recognizing that there are different roles and necessary skill sets for these roles, and of course, you also have the legal and the regulatory aspects and the deployment and operations aspects, right? 

What we have come up with is an AI operationalization framework and actually applying them to projects. So we have developed industrial-specific AI solution blocks, and these serve as foundational elements to build solutions based on AI. This helps us to move to standardized AI offerings by having AI building blocks that are reusable, adaptable and scalable, for example, pre-trained models compliant with the Siemens processes, frameworks, industrial AI services and modules. And we have all the central units working together to operationalize these blocks, like the research and IT units, for example. We also foster co-creation through the ecosystems and partnerships with startups, different companies and research institutes. We’ve set up guidelines and processes, for example, for safety-critical applications based on AI. And there are trainings and communities we’ve set up with the company. So these are some of the ways we’ve tried to overcome the operationalization challenges, and it’s definitely proving to be successful.

HANNAH LABELLE

I’d love to talk about a project example. Is there a specific AI project that you managed that you could walk us through? What was the project’s goal?

SANJUKTA GHOSH

This specific project was about creating digital twins or intelligent objects for assets in the process industry via the engineering documentation. The process industry could mean different verticals like [the] food and beverage industry or a power plant. Let’s say you have the engineering documentation, like piping and instrumentation diagrams or electrical diagrams or electrical schematics.

If you look at these different data sources, we want to then represent in the digital world what a power plant has. So what kind of pump is there? What kind of compressor is there? What are the vessels? What are the pipelines? What are the different tagging systems that are used for these different industrial assets? Because eventually, let’s say, something goes wrong at the power plant, then you need a maintenance person to really go in there and order a part and maybe repair it. So you have a whole bunch of different use cases that kind of follow around the basic premise of at least first having a digital twin. And once the digital twin exists, of course, it leads to more value-added services that one could undertake. 

HANNAH LABELLE 

What sort of support did you have from executive leaders and stakeholders? How did you build buy-in for creating these digital twins with data and AI?

SANJUKTA GHOSH 

Executing such projects needs an interdisciplinary team. So all the skills may not be available now within a single unit. There was a strong support from executive leaders to have the required experts available from across the organization, spanning different units, and to really set up such an interdisciplinary team. Moreover, just the backing and support from the leaders to create a conducive environment in the organization for pursuing such projects where, as we know, AI with all its challenges, risks, also opportunities are involved. And in terms of buy-in, definitely, a buy-in was required for the project from potential users of the system, I would say, and the end customer. It involved really now setting realistic expectations while explaining the potential of using AI.

HANNAH LABELLE

What risks did you have to mitigate and manage throughout the project? 

SANJUKTA GHOSH 

The fact that it’s experimental for one, because you don’t really know which model would work well. So there’s a lot of experiment involved, and then you figure out what would work eventually. So you duly go in cycles and iterations, and you also need to do a continuous validation and testing because you deploy the model and you start using it and then you realize that something is changing or you’re getting some new data. These are the major risks, I would say, and projects can really just slip out of control if you are not wary of these upfront somehow. It’s also about getting the right kinds of data because you need to have the right amount of data and also the right annotations for the data. 

HANNAH LABELLE 

What about collaboration? You had this interdisciplinary team with members who have all different skill sets. How did you get everyone on the same page about the vision and collaborate effectively on a project that had a lot of complexity?

SANJUKTA GHOSH 

Definitely a lot of communication, right? And this is actually even a pretty huge team. The way we structured the project is in different work streams, and you have work stream leads. There’s a lot of collaboration across these different work streams as well. And of course, there are clear roles and responsibilities defined such that the work stream leads are then coordinating within their work streams and also playing the role of coordinating across the work streams, right? Otherwise, there’s easily the risk that in such a large project, people are just not aware [of] what is happening in another corner of the project. I think we’ve set up the structure in this manner such that we make sure that there is as much coordination as possible.

HANNAH LABELLE 

What was the outcome of that project? And how did you measure success with it?

SANJUKTA GHOSH 

We successfully developed a solution for converting these assets represented in engineering documents to intelligent entities, setting the base for the creation of digital twins. The measure of success was that we were able to bring in automation in the process of digitalization of these engineering documentation[s] by successfully using AI. So the concrete measure was the reduction of manual digitalization effort because traditionally, these digitalization[s] had been done manually. So the amount of effort we managed to save in the manual digitalization effort was one concrete metric. And other than that, the ability to generalize and scale, because now having automated the process, one could really scale and even generalize. So the other measure also was about applying what we had developed across other businesses of Siemens in other domains, for example, and not just in the process industry.

HANNAH LABELLE 

How do you see AI projects affecting project managers and other individuals in the future, whether that comes from the skills area or the knowledge base, different things like that?

SANJUKTA GHOSH 

The pace at which AI is evolving is very fast, right? And there are new developments all the time. To keep pace with these developments can be quite challenging while still trying to plan and execute projects that develop or use AI. So project managers will need to find this balance—on the one hand, planning and executing your project, and at the same time looking at this pace of development because, for example, you decide to use a specific model or a tool or a framework and voila, after three weeks there’s something new. So how do you really keep this balance?

Also keeping abreast with new technologies and skill sets, on the one hand, as developers, managers of AI systems, and on the other hand, as users of AI systems, if you will. Having an open mindset to embrace change in the way things have been run traditionally, I think this is something where it really would affect not just project managers, but also other individuals in the future. If you think about it, we’re living in [a] really complex world, and AI does hold the potential to address some of the most pressing challenges of our time, but we will need to harness this immense transformative power with responsibility and [by] keeping things more human-centric.

HANNAH LABELLE 

Sanjukta, thank you so much for talking with me today. And thanks so much for sharing all your insights.

SANJUKTA GHOSH 

Wonderful. I had a great time. Thanks for having me.

STEVE HENDERSHOT 

Thanks for listening to Projectified^®. If you like what you heard, you can listen to more episodes on your preferred podcast platform or visit PMI.org/podcast. And please subscribe to the show and leave a rating or review—it’s always great to hear from you. Hope you can join us next episode!