Companies rushing to embrace GenAI must look—and plan—before they leap. By following these five steps, businesses can unlock the transformative potential of GenAI to drive greater innovation, efficiency, and growth.
Whether they’re using it to hyper-automate routine tasks or to supercharge products and services, businesses are increasingly embracing generative artificial intelligence (GenAI) to gain a competitive edge. But successfully implementing GenAI is easier said than done. It requires a well-defined strategy and a deep understanding of your business needs and data requirements.
The first question businesses should ask themselves is which areas within their organization would benefit most from GenAI. Should initial GenAI projects be focused on improving internal operations, such as making the marketing or HR department more efficient? Or should they be customer-facing, such as optimizing the call center or enhancing a company’s overall digital presence?
Whatever the use case, as a first step, GenAI can be very effective in optimizing and streamlining complex business processes and workflows. In industries like healthcare, technology, manufacturing, and retail, there are intricate procedures spanning multiple departments and agencies. Companies are now starting to identify places where AI content generation, analysis, and summarization can automate handoffs between teams and accelerate timelines and outputs.
One customer we work with, a large healthcare provider that manages the scheduling of radiology appointments for hospitals, has been able to drive massive efficiency gains with AI. Scheduling a single radiology appointment involves evaluating dozens of different parameters, like physician, equipment, and patient availability—a process that previously took 8 to 10 minutes on average. By using GenAI to rapidly analyze all the data points, the provider has cut scheduling time down to just 2 to 3 minutes. That’s big. For a high-volume provider, reducing scheduling time by even a few minutes per appointment translates into millions of dollars in cost savings annually.
But the reality is that most organizations will need to learn to walk with GenAI before they can run. More than a decade ago, consulting firm McKinsey broke down the stages of innovation into three different horizons. The first, which occurs over the initial one or two years, is incremental innovation, which involves making gradual improvements or optimizations to existing products, services, or processes. The second horizon, which occurs over the next two to five years, is exploring and discovering new expansions. The third horizon, typically five to 10 years out, involves envisioning and creating completely new business models or market opportunities that have not existed before.
I believe these innovation horizons still hold true, but their timeframes will be greatly accelerated in the era of GenAI. In fact, according to a recent report from IDC, 92% of AI deployments take 12 months or less, with organizations realizing an average return on their investments within 14 months.
See also: Deriving Layers of Value from GenAI Applications
5 steps to best use GenAI for business
While the potential benefits of GenAI are huge, successfully operationalizing and scaling the technology requires a pragmatic approach with clear goals. Here are five steps to achieve positive outcomes for your GenAI rollout.
1: Ensure AI accuracy
A critical challenge businesses face is verifying the accuracy and reliability of AI-generated outputs. Most AI models, like ChatGPT, come with disclaimers that their content may be inaccurate or wrong. That’s why companies still need humans in the loop to review and validate the accuracy of AI outputs.
The good news is that there are new tools in the market that are allowing humans to evaluate AI-generated content, provide feedback on what is accurate vs. inaccurate, and continuously fine-tune models. Having a degree of human oversight will be essential for building trust and accountability as GenAI gets operationalized across business workflows.
2: Quantify the business case
Another key consideration is ROI. You have to look at whether you should put energy into doing something with AI that could otherwise be done through traditional automation or existing workflows. How much money are you spending to achieve something that could be done in a simpler way?
Beyond financial ROI, companies should map out the potential time savings from using AI for content generation, analysis, and other tasks. If you want to generate and analyze content using AI, how many hours would you save vs. doing it manually? What is the total reduction in time for that business process? You can then map those time savings to cost savings. The time and cost savings enabled by streamlining processes with AI can be compelling drivers. However, those savings can vary greatly across use cases and domains.
3: Find an experienced partner
The rapid evolution of GenAI over the past 18 months has understandably caused confusion. Initially, there was talk of building custom models in-house. But that’s akin to creating your own mobile app platform. It’s clearly an overly complex and costly endeavor for most companies.
The wiser approach for most organizations is to run proof-of-value (POV) projects. Unlike a proof-of-concept that validates the technology itself, a POV demonstrates the concrete value your business can derive by leveraging GenAI for specific use cases. Don’t get bogged down in proving what’s already established—that these models work. Focus instead on proving their value to your operations.
To run an effective POV, tap into the expertise of partners and IT service providers with deep GenAI domain knowledge backed by proven platforms like Microsoft, Google, or AWS. These specialists bring with them valuable experience in implementing similar solutions for other clients.
4: Prioritize data quality
The key to great GenAI outcomes is data preparation. It’s all about the quality, availability, organization, and governance of your data rather than just the training process itself. The bottom line is that it’s impossible to achieve success with AI if your data is of poor quality or unavailable.
Many CXOs remain skeptical about how GenAI can work effectively with their data, given that large language models are typically pretrained on internet data. This notion needs to be demystified. AI models can operate well on an organization’s data as long as it’s correctly structured. The important part is having high-quality, well-organized data at the ready.
5: Implement AI guardrails
GenAI can be unpredictable. It can reinforce biases, jeopardize privacy, and lead to unethical decisions. This is one of the most formidable impediments to widespread adoption. But the challenge can be overcome if ethical considerations, data bias, and similar issues are carefully addressed. For instance, the much-discussed issue of unexpected and unwanted speech. There are now tools that provide the means to control and filter speech related to hate, violence, or self-harm. Such tools are pivotal, and they will instill confidence in users.
In particular, major tech vendors like Microsoft, Google, and AWS, as well as many startups, are developing tools and add-ons to help deploy GenAI solutions with guardrails already in place. The goal of these technologies is to ensure that users worry less about filtering biased outputs and blocking abusive content. That task is handled behind the scenes. But AI safety tools are still in their early stages. Implementing robust guardrails will require significant effort for any business looking to capitalize on GenAI.
See also: Beyond Buzzwords: A Deeper Look into GenAI
Final takeaway
GenAI is now transforming business in real time. And organizations that don’t get onboard are being left behind. But companies rushing to embrace GenAI must look—and plan—before they leap. By following these five steps and staying up to date with the latest trends and best practices, businesses can unlock the transformative potential of GenAI to drive greater innovation, efficiency, and growth.
