Guide to scaling generative AI for your business
Generative AI offers many advantages over conventional AI, but the greatest might be this: It can scale far more quickly. Here at PwC, where we recently rolled out a three-year, $1 billion investment in generative AI (GenAI), we’re already starting to see productivity surge in some areas by as much as 40%. The reason? GenAI is helping us transform high-volume work in processes across our firm, which grows capacity for our teams and provides faster access to relevant data and insights that enhance higher-value work too.
Given the potential, the question isn’t if you will adopt GenAI everywhere — because you likely will, even if you do nothing. Common business applications and platforms such as those offered by Microsoft, Salesforce, Amazon and Google are increasingly embedding GenAI to enhance their user experience. Now the question is whether you’ll use GenAI to transform the way you work more quickly and effectively than your competition. If you do, you can see productivity surge and new business models emerge while you manage the risks and build trust among your stakeholders.
Based on our experience internally and with clients, we’ve developed some guidelines for deploying GenAI successfully. These can help you achieve ROI today, transformation tomorrow and trust in your GenAI security and output every step of the way.
Start with trust and a focus on data governance and security
The place to start with GenAI — as with almost any technology — is by laying the foundation for trust in its design, its function and how its outputs are used. An overall approach toward trusted AI begins with governance, but concentrating on data governance and specific considerations around security is especially important. A secure GenAI system is usually a private one — based on a large language model you have licenses to and operate within your network, surrounded with guardrails and governance, guided by responsible AI practices, to help safeguard data and intellectual property. GenAI-specific governance and usage policies can also help manage potential risks from GenAI capabilities embedded in enterprise resource planning (ERP), customer relationship management (CRM) and other enterprise applications.
To add security to GenAI systems, you may need to look beyond the systems themselves. Assess your network architecture, security policies, data governance and compliance framework in the light of generative AI’s new risks. Don’t forget to consider third-party risk management and ongoing monitoring of how their governance and risk practices interact with yours. After your assessment is complete, close the gaps, whether in security policies or the security environment. In our experience with clients, assessing and enhancing the security environment to protect GenAI can typically be done within 60 days.
Craft a generative AI strategy to quickly realize ROI
It’s usually best to think big here. Look for GenAI “patterns” that can scale and deliver ROI across your company. There are simply too many use cases across every business for GenAI. Instead, build out patterns that can be applied across your processes. For example, generative AI’s capacity for deep retrieval — extracting actionable insights from unstructured data — may deliver modest value in a single area. But if you roll out deep retrieval in every line of business and every function, the ROI can be spectacular. It’s one of the key differences between GenAI and conventional AI, which usually requires a new model for each task. You can often deploy the same GenAI model in many areas quickly.
For even greater ROI, focus on your core business processes. You may find, for example, that the greatest impact on your bottom line won’t come by starting with GenAI for deep retrieval. Instead, you might start with GenAI to write or proofread software, or to customize a customer experience. Naturally, your strategy should also consider the readiness of your data, processes and people — and lay out a plan to close gaps.
Putting together an effective generative AI strategy shouldn’t be too lengthy of a process. Think in terms of weeks, not months, knowing it will continue to evolve.
Commit to use cases — and launch
Your initial pilots should follow your strategy, with its focus on core business processes, organizational readiness and repeatability. A use case of moderate value that you can quickly replicate may deliver more ROI than a high-value one-and-done use case. Since GenAI models come pre-trained, you should be able to stand up initial use cases with a 90-day sprint.
To get these launches off the ground, you’ll likely need a core GenAI team that includes not only technology specialists, but also analysts and leaders from the business. It’s your business professionals who know what your company needs, what differentiates you in the marketplace and who will identify the data and provide the insights needed to customize GenAI models. GenAI can then produce relevant, accurate outputs — which can eventually enable all your people to access the expertise of your most experienced professionals.
Discover more generative AI insights
Generative AI is here
Explore how it can reinvent the way we work, accelerate innovation and drive trusted outcomes.
Stand up an AI factory — for fast, repeatable, verifiable results
Your first use cases are now up and running. Your core team is noting any issues, so you can close gaps in data, technology and skills. They’re also identifying successes for you to replicate with an AI factory. A generative AI factory features a series of pods, each focused on a different domain or line of business, and each with combined technology and business expertise. At PwC, each of our AI factory pods 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 customize 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. At PwC, we have pre-built GenAI toolkits to support personalized customer experiences, content creation, research, agile software delivery, support services, report generation, deep retrieval, smart summaries, Q&A engines — the list goes on.
Transform your workforce and grow ROI further
Once your generative AI factory is in production mode, your focus should shift to integrating more of your data into GenAI and transforming processes with it. The goal is for your entire workforce to benefit from GenAI by making it an integral part of major processes and transforming how people work faster and more effectively. That requires a top-notch user experience to use GenAI directly and as embedded capabilities in your enterprise applications. It also requires upskilling people on responsible use of AI.
To safeguard your systems and grow ROI further, continually assess generative AI’s outputs, costs, performance and alignment with business objectives and risks.
This effort should start with a responsible AI framework that covers strategy (for the CEO and board), controls (for chief risk and compliance officers or others responsible for the implementation of controls responding to the identified risks), responsible practices (for chief information and information security officers) and core practices (for the AI factory, including data scientists and business analysts). It also includes comparisons with historical data sets to help validate outputs and periodic audits by teams that include both domain specialists and data scientists as well as those with the right objective perspective on whether the organization’s trust objectives are being achieved.
The GenAI difference: Transformation can happen quickly
In our experience deploying GenAI both internally and with clients, the same phenomenon occurs again and again. Scale takes place and transformation begins far more quickly than those new to GenAI expected.
The key to rapid ROI and far-reaching transformation through GenAI is a focus on discipline, scale and trust — as the five guidelines above outline. Secure your environment. Craft a strategy. Choose the right place to start and close technology, skills and data gaps. Build an AI factory to speed deployment. Establish rigorous oversight that includes responsible AI. With these steps, you can quickly achieve a surge in productivity and build the foundation for new business models based on hyper-personalization, continuous access to data and insights and more.
Generative AI
Lead with trust to drive sustained outcomes and transform the future of your business.
What can generative AI do for you?
Generative AI is already transforming business. Contact us to learn more about this rapidly evolving technology — and how you can begin putting it to work in a responsible way.
Next and previous component will go here
