How to achieve AI maturity: Tips from experts at AWS and Fannie Mae

75% of leaders have paused or delayed AI projects. To make sure AI initiatives succeed and drive ROI, organizations need to boost their AI maturity.

Drew Firment, Pluralsight’s VP of Enterprise Strategy; Chris Hennesey, Enterprise Finance Strategist at AWS; and Daniel Seeley, VP of Cloud and Infrastructure Engineering at Fannie Mae, share how organizations can make the most of new AI services and achieve AI readiness.

Watch the webinar now.

The AI hype calls to mind the cloud computing buzz from more than a decade ago. But the speed of these two digital transformations is vastly different. Cloud evolved gradually along a predictable path, typically taking five or more years to implement. AI, on the other hand, has undergone rapid, unpredictable acceleration. 

This has led some organizations to explore AI for fear of missing out (FOMO). But more strategic organizations are methodically experimenting with AI to understand what it could bring to their business.

Fannie Mae is one such company. “[AI adoption] is a ‘no regrets’ move that will ultimately enable Fannie Mae to reallocate some money for us to continue to test and learn where AI is going to have an impact on our business. We’re also introducing some technology in the spirit of developer and overall user productivity,” says Daniel.

With organizations spending up to 10% of their technology budgets on generative AI, the more strategic approach is an indicator of AI readiness—and the one more likely to drive ROI.

With just half of AI projects making it into production, it’s no surprise that only 9% of organizations are considered AI mature. But what does that really mean? 

Drew, Chris, and Daniel break down AI maturity into four key themes:

• A robust cloud foundation
• Security practices to ensure safe infrastructure and data management
• A structured data strategy
• Ongoing skill development  

They also identified three phases of AI readiness: 

1. Tactical (crawl)
2. Strategic (walk)
3. Transformational (run)

So how do you actually achieve AI maturity across those key themes and reach the transformational stage? Drew, Chris, and Daniel share their insights.

Just as successful cloud adoption requires best practices, AI adoption demands structured approaches to using models and ensuring compliance. But the premise of making sure you have 100% of the requirements upfront before starting an AI project doesn’t make sense in practice. 

“Failing fast is actually encouraged,” says Daniel. “You don’t need 100% of your requirements to get started. Start today and come back in two weeks at the end of a sprint.”

In some environments, especially large companies, people don’t have insight into the cause and effect of their roles. Make sure people who are consuming AI understand the costs and benefits associated with it. 

“Always federate as much as you can about cost insights to the consumers so they can understand and make necessary trade-offs,” says Chris. 

It’s also important for them to understand the value of the technology, such as the potential to scale revenue or new products and services in an infinite way. 

Identify areas where AI can have an impact in your organization. For example, Fannie Mae is exploring the use of AI to help engineers ingest VPC flow log data and more easily monitor network traffic for security and troubleshooting purposes. 

When possible, establish KPIs aligned with your AI use cases. But understand that you may not be able to quantify every impact.

Chris says, “It’s great to quantify things within generative AI, but there are a lot of unknowns around how value will be delivered and what products or services will be created. Be open to not having to quantify everything as you go through this process. At this point, underinvesting and falling behind your competition is a greater risk than overspending on AI.”

Only 29% of organizations have dedicated resources to manage AI risks. Develop a governance framework for AI that includes:

• Operational control
• Cost control
• Data privacy
• Ethical AI policies 
• InfoSec requirements
• Intellectual property management
• Risk management oversight
• Usage monitoring

The key is to provide a controlled, secure environment without limiting the creativity of data scientists and other teams working to develop valuable solutions.

Learn more about establishing AI governance.

“You’re not going to get as much out of AI if you don’t have your data in place,” says Drew. 

To succeed with AI long term, it’s critical to know:

• Where is your data?
• Which data stores contain sensitive data?
• How will you store the output and control access?
• Can you support diverse data types and low latency storage?
• Are you using data pipelines for real-time data ingestion?
• Are you leveraging APIs?

Learn more about preparing data for AI.

An early indicator of AI success is knowing what data is being pulled into your models and monitoring the end product. While this may seem basic, these controls allow organizations to experiment with certain data until the ecosystem is ready to explore more use cases.

When it’s time to actually run your AI systems, look at performance metrics like elasticity, resiliency, and load balancing. These help you ensure smooth transitions and optimize infrastructure when integrating AI into production environments.

AI requires a culture shift and change management. Leaders need to address the potential fear, uncertainty, and doubt about their organization’s AI strategy and adoption.

They can uncover these concerns by posing these three questions to everyone at the organization:

• Why do you want to explore AI? The why needs to be clear. For generative AI, it could be to create new businesses or boost productivity.
• How do you plan to implement and assess AI? Each company will have a unique approach and strategy.
• What does AI mean for me? This question empowers everyone to use AI.

Education is key to improving adoption and reducing resistance. The more informed people are, the more easily they will adopt and adapt. That’s why upskilling everyone in AI is so important to AI maturity. 

”Everyone understands the potential business value of artificial intelligence, so I can't say enough about the requirement to just continue to learn, learn, and learn some more,” says Daniel. “Infrastructure is no longer a competitive advantage—it’s a commodity. The differentiator now is skills.”

Success varies for each phase of the AI maturity journey. At the tactical phase, organizations are just getting started. Then, as they evolve to the strategic and transformational phases, success looks much different. 

To find out where your organization is, complete our 12-question AI Readiness Index. You’ll see how you stack up against your industry peers and get recommendations to boost your organization's AI maturity based on your results.

To get all of Drew, Chris, and Daniel’s insights, watch the on-demand webinar.