Five tips for AI implementation success

With more than 80% of AI projects failing to deliver on their promise, healthcare leaders face a critical question: how do you ensure your AI initiatives succeed where so many others fall short?

At Noteworthy 2025, Notable’s annual summit on AI transformation in healthcare, a panel of healthcare innovators and leaders came together to discuss what it takes to implement AI into healthcare operations successfully. Moderated by Tom Foerster, Head of Platform Delivery at Notable, the session featured insights from Meg Dingae, Director of Human Experience at Montage Health; Daniel Thomas, VP of Information Technology Operations at Regional One Health; Gabriella Rozenblat, Senior Manager and AI Platform Architect at Notable; and Emily Colson, Team Lead for Customer Success at Notable.  

Their conversation highlighted the critical steps, from setting clear governance and project charters, to deeply understanding current-state processes, and scaling with intention. Here are their top tips:

1. Identify your north stars: Governance and project charters

Before embarking on an AI implementation project, it’s vital to ensure everyone is on the same page. 

Emily Colson identified establishing a governance process as one of the greatest indicators of success for an implementation. It’s important to identify the key decision makers and establish a regular cadence for meetings to ensure alignment on the right priorities at the right time.  

Another important method for success is to set a project charter from the start. “That’s our North Star,” said Colson. “We all have things we’re trying to achieve. A project charter takes the emotion out of it. It can help reset and make sure we’re doing what it says, or maybe not doing what it doesn’t say, and then move forward.”

Meg Dingae agreed, adding that a project charter ensures that they’re “bringing everyone along on the story” at Montage. It allows for consistent review of metrics, and gets everyone excited about the project.

2. Align early on goals and metrics 

From there, the next step is to align early on goals and metrics. Gabriella Rozenblat emphasized the importance of addressing questions such as “What does success look like?” 

Daniel Thomas added that when he and his team set the goals and the metrics up front, “we know what we’re trying to achieve and what things we’re going to measure, and we’re holding ourselves accountable to those plans. It makes it a whole lot easier.”

When setting these metrics, Dingae added that it’s important to involve both leaders and operational team members: 

“We don’t do it top-down, but I think the importance of having the executive sponsors agree that you’re working on the right thing is really important. So you want to have the operational folks on board, but you also want to have your executive team saying this is an important thing to do.”

3. Know your current processes before you automate them 

Once you’ve aligned on your project charter, goals, and metrics, both Thomas and Dingae warned against jumping into automating processes without first knowing them deeply. 

“Do not assume you know what’s going on in current state,” said Thomas. “We found out, week after week after week, that an organization where the registration and check-in process should be the same no matter what desk you go to, was different at every desk you went to. We were constantly having to dig through that.”

“Knowing current state is going to help a great deal moving forward,” he added. 

Dingae stated that establishing standard processes is actually another selling point for executive buy-in. 

“Sometimes it can be hard to sell the workflows that you want to do across the organization. But I think because there’s so much variation in current state, talking about standardization is kind of sexy to the powers that be,” she said. “Part of what you’re doing here is developing a workflow that can be standardized across your organization to avoid duplication and inefficiencies, because we definitely found that.”

4. Start small and then scale  

While it may seem enticing to launch all of your workflows at once for maximum impact, starting small is actually the way to go for a successful implementation. 

“I always tell folks to start small, look for the operators in your organization who are kind of like innovators. They’re not afraid to fail a little bit or not afraid to iterate and try something different,” explained Dingae. 

For example, if you’re automating referrals, start with one pilot site or department. Test, iterate, and perfect the flow before launching it across other areas of the organization. 

When it’s time to scale, Dingae recommended looking at different data inputs for problem areas: 

“We use patient experience data to see where the most friction is. We look for areas that really need help and no one’s actually able to do the work, like care gaps for example.”

Thomas added that Regional One “started small and stayed small for a long time.” However, once they had a workflow that they were sure was right, they were able to rapidly scale the solution to the rest of the organization over a time period of just six weeks. 

“Now we’ve got momentum, so we can take the next piece and move it much quicker down the road,” he explained.

Rozenblat added that having a value roadmap helps everyone align on how, when, and where to scale in a project. She recommended addressing the different challenges the organization faces, ordering them in a priority list, and walking through the described impact and value to make informed decisions about what to focus on. This also helps work towards the longer-term vision. 

5. Make sure you’re working with the right partner 

Having a vendor that you see as a partner on your AI journey is vital for success. Emily explained why:

“We are there with you along your journey. And regardless of what’s going on, we are that person that you can reach out to.”

Thomas shared that working with a committed partner like Notable is different from working with any other vendors because the two teams have put in the time to align on goals and have tough conversations when something isn’t working out. “We've had several [tough conversations] through the implementation on both sides. And it's never been bad, it's never been ugly, the answer has never been “no” from either side.”

Dingae added that accessibility to help makes a big difference, and that she appreciates having  access to “an army of people who can help solve problems.”

A path to successful AI adoption

The path to successful AI adoption in healthcare isn’t always linear, but with the right foundation, mindset, and partnership, it can drive transformative outcomes. The experiences shared by these leaders underscore that strategic alignment, operational insight, and thoughtful scaling are key. 

And perhaps most importantly, having a trusted partner who walks alongside you, helping navigate challenges and celebrate wins, can make all the difference.

Explore more insights from Noteworthy 2025 here.