How PwC accelerated its GenAI journey to help clients reach value at speed

View Transcript

Investing in Artificial Intelligence, above any other emerging tech, is being prioritised by our clients. Since the hype around GenerativeAI (GenAI) took off in earnest around 2022, clients have sought guidance around how to run transformative GenAI programmes. But excitement around the technology took off at a speed that left real world examples far behind - the kind of examples which could help shape best practice.

That's why we decided to become our own 'client zero'. We've had a dedicated AI function for around a decade now and more recently we moved to deploy the technology widely across the firm, exploring use cases and learning lessons that we could share with our clients, helping them get value from GenAI at speed.

“We drank our own medicine,” says Bivek Sharma, Chief AI Officer at PwC UK, of the decision to become our own client zero. “Because the market is so new we couldn’t point to a whole number of clients and say we’ve taken them through the entire journey and here’s the output.”

We have started to transform our operations to take advantage of the opportunities GenAI offers, improving productivity and efficiency, generating new revenue, and future-proofing the business. Along the way we've learned lessons not just about the technology and infrastructure required, but about the skills and cultural change needed to underpin success.

Euan Cameron, AI & Emerging Tech Research and Development Leader at PwC UK, says: “With our dedicated AI function in place, we were already an early mover, so when the latest version of GenAI landed we focused quickly on developing our own secure environments where we could deploy these tools and experiment with them with real data to really understand the potential.”

The route to value

Umang Paw, Chief Technology Officer at PwC UK, says: “It was really important to be client zero. It allowed us to build the experience of how AI can support our clients, both from a sector perspective, but also from a functional perspective.”

To explore GenAI's potential, we ran a series of pilots with over 500 staff, identifying more than 1,000 use cases across multiple GenAI platforms and models. The results pointed to efficiency gains of up to 86% in content summarisation and accuracy levels of around 80% in translations - with employees saving an average of six hours per week on certain tasks.

Today, more than 15,000 of our UK staff are equipped with AI skills and tools. As an early adopter of AI and automation technologies, we've forged alliances with other industry leaders including Amazon Web Services (AWS), Microsoft, Google Cloud Platform (GCP), OpenAI and Salesforce, and we collaborate with more than 20 academic and research institutions around the globe. This network allows us to assess the full landscape of available solutions, and develop a deeper understanding of when to build and when to buy. We learned that while some use cases need proprietary solutions, off-the-shelf options can sometimes be the quickest route to value.

“There’s a lot of people in the market saying, ‘build it yourself, do proprietary’,” says Sharma. “Actually, one of the things we learned is when to build and when to buy, and we know the right platforms to use to get to value quicker.”

Maintaining momentum

Transformation is not a one-and-done project. To be successful, the search for and implementation of better ways of working must never stop. Guided by a human-led, tech-powered approach to innovation, we developed a flywheel methodology to help realise value from GenAI initiatives.

The iterative process begins by assessing the potential value, risks and rewards of GenAI implementation and then developing use cases and tools. We work with our technology alliance partners, alongside our clients, to activate GenAI within our client's technology ecosystems or to build unique solutions. At the same time, we evaluate costs and carbon impacts before deploying, learning from and scaling solutions.

“It has been fascinating to look at the rollout of GenAI at the pace that we’ve been able to move,” says Peggy Gondo, Chief Information Officer for PwC UK. “Already 80% to 90% of our people have access to a GenAI tool and are starting to use that in their day-to-day roles, which is really quite exciting and transformative.”

At every step of our own GenAI journey we have developed insights and learnings that now benefit our clients and support their GenAI success. “The amount that we’ve learned from the use cases we’ve delivered, the infrastructure we have to put in place, how to look at the business value, where to focus, is invaluable in helping us accelerate our clients’ journeys,” says Sharma.

To learn more about how PwC can guide you on the road to GenAI success, please get in touch.

Bivek Sharma
If you look at what clients are looking for right now, it's guidance around how to run these AI transformation programmes. They're looking for people who've done it.

Now, because the market is so new, we can't as we do traditionally point to a whole number of clients and say we've taken them through the entire journey, here's the output.

So using ourselves as a credential, we've drunk our own medicine. We've shown the value.

Umang Paw
It was really important for me to be the first client - Client Zero.

Fundamentally it allows us to build the muscle, allows us to build the experience of how AI can support our clients, both from an industry perspective but also from a functional perspective as well.

Peggy Gondo
It has been fascinating to look at the pace that we've been able to move and already probably 80 to 90% of our people have access to a GenAI tool and are starting to use that in their day-to-day roles.

Euan Cameron
We've had a dedicated AI function in the organisation for almost ten years now, with professional data scientists in it who are keeping their eye on developments within this domain.

So when the latest generation of GenAI landed in late 2022, although we've been working with large language models for some time, it was clear that this new wave was going to have a pretty significant impact on our business and our workforce.

But at the same time, we focused on quickly developing our own secure environments where we could deploy these tools and experiment with them with real data to really understand the potential for the things that we do.

Peggy Gondo
We have a real track record over the last, probably seven or eight years of democratised technology use. So people are used to the concept of applying that in their roles and using the tools as part of the way they deliver their services.

Bivek Sharma
So the other really important thing is the amount that we've learnt from the use cases we’ve delivered, the infrastructure we need to put in place, how to look at the business value, where to focus first, so we can accelerate our clients’ journey on the back of that.

Umang Paw
So we talk about human-led, tech-powered, and actually it really fits in well with generative AI because you have to have a human in the loop.

At least where the technology is today, the importance of a human actually working with the AI, but most importantly, checking what the AI is delivering is absolutely critical.

Bivek Sharma
This is a change journey as much as it is a technology journey for our clients. What we found through our own Client Zero story is the upskilling, the learning and development pathways, the change management needed - is just as important as the technology estate itself.

Peggy Gondo
This is one of those technologies that continues to grow, evolve new functionality, new capability, keeps coming out.

But what I think we've set ourselves up for is the ability to move at pace, to react to the new functionality, the new capabilities in the market quickly.

Bivek Sharma
There's a lot of people in the market saying, well, build it yourself, do proprietary. And actually one of the things we understand because of how closely attached we are to the AI alliance community, is we know when to build, but when to buy. And we know the right platforms to use to get to that value quicker.

And I think there's our pragmatism as well as how quickly we can get to that value, and also our understanding of all the functional domains, which is where all the value sits, whether that's margin or whether that's top line.

Umang Paw
Organisations shouldn't be stalling on generative AI, OK. Actually being first, but also, starting to build that muscle around generative AI, just like we have been doing at PwC, is going to be so important for the future.