Generative AI success: Insights from AWS customer journeys

In 2023, we opened the AWS Generative AI Innovation Center to help customers turn AI potential into real business value by connecting them with our network of AI scientists, strategists, and engineers. Today, thousands of organizations—from small startups to large enterprises—are getting smart guidance and deploying AI applications seamlessly and with confidence.

Recently, an MIT study claimed that 95% of GenAI initiatives fail—that means only 5% succeeded. While their analysis of the challenges is insightful, that failure rate doesn't match the Amazon Web Services (AWS) experience.

This year at the AWS Generative AI Innovation Center 65% of customer initiatives moved from proof of concept into production—some ready in as little as 45 days. From the PGA TOUR to Rocket Close to Experity and more, here’s how we’re setting organizations up for success. 

1. Like everything we do at Amazon, we start with the business outcome and work backwards. Our team collaborates with the customer to clearly define business objectives on day one. We keep this production-focused mindset as we build roadmaps and develop solutions with customers.
2. We embrace agility as a competitive edge and plan for change. Organizations that thrive won’t be those who perfectly predict the future, but those who adapt quickly as it unfolds. Customers that achieve production have created applications that adapt to changes in models and hardware. Whether teams leverage “off-the-shelf” models, or build their own (or both), for their systems to scale, the applications need to be nimble as new models. 
3. Our experience with thousands of customers has revealed a clear pattern: the most successful AI implementations start with a strong data and cloud foundation. Know that data is your differentiator—combining models with your organization’s proprietary data, transforms AI into powerful, business-specific applications. This allows you to create customized customer experiences, such as highly personalized customer service agents that can help resolve issues faster. The PGA TOUR is a great example of this—using Amazon Nova, the organization achieved 70% cost savings compared to alternative AI models, all while maintaining its high editorial standards
4. Security must be built in on day zero. Customers move fast when they have the right controls and data protections established from the start. We map out your application’s intended actions to develop a secure and responsible application. Adding guardrails and other security measures afterwards invites more risk, which can cause organizations to lose trust with their customers. 
5. When it comes to building scalable and efficient AI applications, discipline equals freedom. Before we get to the fun of launching an application into production we must focus on operations, governance, and maintaining solutions that scale. Having this foundation upfront allows customers to adapt to the changing needs of their businesses. 
6. Prepare teams to think beyond the technology today. Organizations that are successful on their AI journeys are those that invite their employees to be a part of their business transformations. We partner with customers to create opportunities so their teams can learn more about their AI applications through our upskilling and training programs. The team at Experity had a long-term vision in mind as we worked with them to develop a multi-agent, digital assistant, which has supported more than 30,000 patients across 160 clinics.

With these proven principles, AWS customers aren’t hoping for ROI—they’re achieving it.

Having the right partners is how your organization can get out of the endless experimentation loop and on a roadmap to build smart, secure, and scalable applications.