The buzz around Generative AI (GenAI) is undeniable. Organisations are eager to tap into its transformative potential. However, our own experience tells us that while the enthusiasm for Gen AI is palpable, there lies a significant gap between desire and readiness. PwC’s 27th CEO Survey found the same, with most organisations yet to move beyond GenAI experimentation and small-scale pilots.
While adoption is — rightly — being driven by business leaders, it is being supported and enabled by Technology teams. Often these teams are already at capacity or have not yet developed the necessary skills or resources to deliver and support GenAI. In this article, we share the need to build the right operating and delivery model for GenAI to unlock its full potential, how to use it within the Technology team itself and how we at PwC approached it.
Build the right operating model
How can the performance of GenAI tools be maintained once embedded?
While it's unnecessary to overhaul existing operations for GenAI, organisations will need to manage increased volumes of monitoring and analysis to prevent and manage GenAI-driven compliance and security breaches. This heightened requirement may accelerate the need for organisations to mature their Tech and Data Operations capability through increased automation. Big data and machine learning are being adopted to automate Tech operations processes — including event correlation, anomaly detection and causality determination.1
Furthermore, Gen AI will also require an evolution of traditional tech ops roles to include more advanced skills, with proficiency in reliability engineering, data science, and security. It is crucial for operations teams to proactively strategise how they will acquire these advanced skills before the need arises.
Build the right delivery model
How can the Technology team best support business functions to deliver?
The delivery model does not need to be a different concept to the broader enterprise delivery model but should ensure there is enough dedicated capacity and the right skills to drive iterative delivery and improvement of use cases.
Here at PwC, we set up our own GenAI Delivery Factory that delivers GenAI solutions to use cases for the business. The factory is a series of pods, each focused on a different domain or line of business, and each with combined technology and business expertise. Each pod contains business analysts, data scientists, data engineers and two GenAI-specific roles — a prompt engineer to refine the GenAI model’s output and a model mechanic to oversee and customise the model’s inner workings.
The pods themselves can scale: sharing executive oversight, governance, user experience designers, data science support and toolkits with reusable software code and prompts.2
How can GenAI transform the way technology is delivered?
Identify, shape and deliver the most valuable GenAI use cases for your Tech teams and deliver on them.
GenAI can be used to augment the Tech workforce and improve speed of delivery, user experience and risk management, enabling employees to focus on more strategic work. This presents a significant opportunity for technology teams looking to drive increased efficiency and greater return on technology investment. Technology teams should be looking to develop their own Use Cases to capitalise on this opportunity. Examples include:
Tech user support: next generation user support through GenAI triaging, resolution of simple, repeatable requests, and troubleshooting through more complex issues.
Architecture blueprinting: GenAI is able to consume large amounts of system usage data to develop recommendations for treatment of applications, freeing up capacity of in-demand architects.
Code development: complete automation of basic coding tasks, offering real-time feedback, and learning from existing code. For example, GenAI can be considered in migrating between code languages or between natural language and code. Tasks such as these, traditionally requiring significant funds and resourcing, become more accessible through GenAI. Using GenAI for code development will change how technologies are delivered and who delivers them, disrupting today’s delivery models.
Data quality management: automating data management processes such as cleansing, data integration and data quality management. These are typically resource-intensive tasks that can commonly delay delivery timelines. Harnessing GenAI for data management will reduce manual data management, driving increased efficiency and certainty in delivery timeframes.
Testing: GenAI can automate test case generation, test execution, library management and even bug-fixes. This capability can support in-demand business and technology specialists commonly resource-constrained in organisations with high volumes of change. GenAI for testing will free up Business and Technology SMEs from laborious testing activities, and talent focus will shift to more strategic work.
How did PwC do it and what did we learn?
We have committed to a three-year, $1 billion investment in GenAI, and we’re already starting to see productivity surge in some areas by as much as 40%. Our plan is to deploy our GenAI to every one of our PwC professionals globally.
- We found focusing on individual use cases (across the entire organisation) was erosive to value. There are simply too many use cases across all lines of business.
- We looked for GenAI ‘patterns’ that could scale and deliver ROI and built out those we could apply across our processes.
- We track ROI through these patterns — as we can then understand who is using them.
- We have pre-built GenAI toolkits to support personalised customer experiences, content creation, research, agile software delivery, support services, report generation, deep retrieval, smart summaries, Q&A engines — the list goes on.
If you would like to talk about how your operating model can enable GenAI impact, contact Matt Benwell.
1 https://www.gartner.com/en/information-technology/glossary/aiops-artificial-intelligence-operations
2 https://www.pwc.com/us/en/tech-effect/ai-analytics/scaling-generative-ai.html
