"Up to 80 to 90% of AI projects fail to deliver. So there's really a huge gap still."

That is the first thing Sarah Cornett said when we sat down. Sarah is the founder and CEO of Global AI Advisors. She helps organizations turn AI strategy into business value, and she comes at it with a background in quantitative finance and enterprise strategy.

So I asked her the obvious question. Why is the number that bad?

Her answer was mostly about people.

"Some of it is technical, like data availability or infrastructure. But one of the biggest reasons for failure is leadership."

The failure starts at the top

Sarah's read is that leaders are still catching up. In her words, "Leaders are still learning how to adopt AI, how to align it with strategy, and how to guide teams through change."

I see the same gap from the other side. Without structure, people are spending money on ChatGPT with no oversight. Licenses get bought. Nobody owns the outcome. Then the project lands in the 80-90%.

The fix Sarah described is boring on purpose. Strategy first, then use cases, then people, then governance. Skip a step and the pilot stalls.

Strategy before tools

Her first piece of advice:

"First, I always advise starting with an AI strategy. What are your goals? Where is your industry going? What value could AI bring to your customers?"

Once you have that, you rank use cases on two things: business value and feasibility. Feasibility means your data is ready, your stack can support it, and the risk is something you can live with.

Sarah was blunt about scope. "You don't want to start with something that could take 10 years to implement." Find the high-impact, low-hanging fruit. That is where momentum comes from.

Pick the project your current data can support this quarter. Prove it works. Stack the next win on top of it.

Train people on their actual job

A generic prompt engineering course will not fix adoption. Sarah has run training for analytics teams and marketing teams, and her conclusion was specific:

"the most powerful training is always customized to the tools, tasks, and challenges that specific team is facing."

Pair that training with safe-use guidelines: what data can go into a tool, what cannot, and how to handle IP and regulated information. People need to understand their roles with AI. Leadership changes too.

Every role lands in one of three buckets

Sarah has spoken at conferences on the future of work, and she sorts the impact into three buckets.

Replaced. Some roles go away, the way memo writers did after the Internet arrived.
Augmented. AI handles the busywork so people can focus on higher-value work.
New. Roles that did not exist a few years ago: AI strategists, ethicists, prompt engineers.

If you do not tell your team which bucket their job is in, fear fills the silence and adoption dies. Sarah put it this way: "We can't just replace people, we need to lift them up and help them adapt."

Governance before production

This is the part most growing companies skip. Sarah has stood up governance programs at banks and healthcare organizations. In regulated industries like healthcare, governance comes first because the regulators require it.

For everyone else, her order of operations is clear:

"If you're new to AI, start with your data governance. Then, establish a governance board, educate them, and build a workflow for approval before AI is pushed to production."

Keep data governance (quality, lineage, access) separate from AI governance (model risk, explainability, approvals). Put legal, compliance, security, data and operations on the board. Watch the EU AI Act and the state-level US policies that apply to you.

I added one piece. People have to understand how the AI came to its decision. That is leadership's responsibility. Governance feeds data, data feeds security, security feeds workflows, and workflows produce business outcomes.

The tools Sarah uses every day

Sarah uses ChatGPT, Claude and Perplexity almost daily. Lately she has been playing with Perplexity's Comet, their version of web browsing. Her words: "it's really powerful."

Tools support the roadmap. They do not write it.

What to do this week

Write a one-page AI strategy: your goals, where your industry is going, and what value AI could bring your customers.
Score your use cases: list every idea on the table and rate each on business value and feasibility. Cut anything that needs years to land.
Pick one pilot: high value, high feasibility, with a named owner, one success metric and a 4-8 week timeline.
Name a governance board: legal, compliance, security, data and operations. Write down the intake, review and approval path before anything ships.
Schedule team-specific training: built on the tools and data that team actually uses, plus clear safe-use rules.
Tell your people which bucket they are in: replaced, augmented or new. Say it early and say it often.
Try Comet: spend twenty minutes researching a real business question with it.

Listen to the full conversation

Sarah and I went deeper on governance, explainability and what human-centered leadership looks like in practice. The full episode is below. If you are trying to get out of the 80-90%, start there.

Connect with Sarah Cornett

Global AI Advisors: https://www.globalaiadvisors.ai

LinkedIn: https://www.linkedin.com/in/sarahcornett-ai/

Watch / Listen to the Episode: https://bry.net/ai

Follow me & join the conversation

Facebook: https://www.facebook.com/profile.php?id=61575757332333
Instagram: https://www.instagram.com/aiwithbry
TikTok: https://www.tiktok.com/@aiwithbry.com
YouTube: https://www.youtube.com/@aiwithbry
X (Twitter): https://x.com/aiwithbry
LinkedIn: https://www.linkedin.com/in/phingers

Episode's Transcript

Please understand that a transcription service provided the transcript below. It undoubtedly contains errors that invariably take place in voice transcriptions.

**[00:00.000] Sarah Cornett:**Up to 80 to 90% of AI projects fail to deliver. So there's really a huge gap still.

[00:23.342] Bryan Dennstedt:

It feels like it's almost going back to the basics, like, get out the crayons and color, and learn how to draw again. There are all these new opportunities now in this AI world that we're in.

**[00:23.342] Bryan Dennstedt:**Hey everybody, welcome back to AI with Bry, the show where we dive deep into how to learn, leverage, and lead with AI. I'm your host, Bryan Dennstedt, technologist, strategist, and someone deeply curious about how AI is reshaping our lives and work.

Today, I'm thrilled to welcome Sarah Cornett to the show. Sarah, welcome.

**[00:39.512] Sarah Cornett:**Thank you, Bryan, so excited to be here.

**[00:41.271] Bryan Dennstedt:**Sarah is the founder and CEO of Global AI Advisors, helping organizations turn AI strategy into real business value. She brings a unique blend of quantitative finance, enterprise strategy, and deep knowledge in governance and AI adoption.

We've gotten to talk a few times about how to make AI practical, impactful, and people-centered. I'm really looking forward to today's conversation because you're bringing both technical depth and business perspective, which is something so many leaders are trying to balance right now.

**[01:09.112] Sarah Cornett:**Absolutely. Yes. We are definitely living in a new AI era. There are so many ways we can talk about the future of AI and what this means for organizations.

**[01:23.580] Bryan Dennstedt:**Let's dive into the Learn segment. What's something you've learned recently that really stood out?

**[01:29.442] Sarah Cornett:**From working with organizations on AI transformation, what really stands out is that we still have a long way to go. I recently learned that 80-90% of AI projects fail to deliver expected value. That was an alarming statistic.

**[02:39.138] Sarah Cornett (continued):**Some of it is technical, like data availability or infrastructure. But one of the biggest reasons for failure is leadership.

Leaders are still learning how to adopt AI, how to align it with strategy, and how to guide teams through change. There's a major gap in education and leadership around AI.

**[03:29.512] Bryan Dennstedt:**It's so true. It feels like we have to go back to the basics, like, touch it, play with it, break it, and see what it can do. You touched on the data gap, but also on workflow. There are so many ways you can apply AI across roles. How do you help leaders learn and experiment?

**[04:57.280] Sarah Cornett:**Great question. First, I always advise starting with an AI strategy. What are your goals? Where is your industry going? What value could AI bring to your customers?

Then you prioritize use cases, not just by business value, but also feasibility. You don't want to start with something that could take 10 years to implement. Find the high-impact, low-hanging fruit, that's where you build momentum.

**[07:10.868] Bryan Dennstedt:**What's a tool you've recently used that everyone should check out?

**[07:15.412] Sarah Cornett:**I use ChatGPT, Claude, and Perplexity almost daily. But lately, I've been playing with Perplexity's Comet feature, it's like their version of web browsing, and it's really powerful. Highly recommend trying it.

**[08:02.300] Bryan Dennstedt:**You've already started touching on this, but let's unpack leverage. I believe AI is going to be like electricity, every business will use it. Where do companies start?

**[09:12.428] Sarah Cornett:**It always starts with your strategy, that becomes your guide. Then, once you've identified the right use cases, you need to layer in education and change management.

People need to understand their roles with AI. How does leadership change? How do teams evolve?

I've worked with analytics teams, marketing teams, and the most powerful training is always customized to the tools, tasks, and challenges that specific team is facing.

**[11:28.098] Bryan Dennstedt:**I love that. And I think as AI becomes more embedded in orgs, planning becomes more important. We're used to agile, but maybe more detailed planning enables better AI results. Are you seeing that?

**[12:55.320] Sarah Cornett:**Totally. It starts with leadership, understanding team structure, AI vs. human responsibility, and defining the workflows.

I've spoken at conferences on the future of work, and we see three big buckets:

AI replaces roles, like memo writers in the pre-Internet era.
AI augments roles, freeing humans to focus on high-value work.
New roles emerge, AI strategists, ethicists, prompt engineers.

**[16:13.228] Bryan Dennstedt:**Yes! And I say this almost every episode: we're entering a new Renaissance era, creativity will explode. But we need structure, especially governance. Without it, people are spending money on ChatGPT with no oversight.

**[17:22.301] Bryan Dennstedt:**Let's close with leadership. How are you helping teams grow in this space?

**[17:32.412] Sarah Cornett:**I'm doing a lot around AI leadership development, especially educating leaders on governance, investment, and human-centered adoption.

Governance is huge. And human-centered leadership is critical. We can't just replace people, we need to lift them upand help them adapt. That's the kind of leadership this moment calls for.

**[20:46.122] Bryan Dennstedt:**Yes, and we need explainability too. People have to understand how the AI came to its decision. That's leadership's responsibility.

**[21:14.372] Sarah Cornett:**Exactly. I've worked with banks and healthcare orgs to stand up governance programs. Some industries, like healthcare, start with governance first, due to regulation.

If you're new to AI, start with your data governance. Then, establish a governance board, educate them, and build a workflow for approval before AI is pushed to production.

**[23:04.992] Sarah Cornett (continued):**Make sure you're aware of regulations like EU AI Act or state-specific US policies. Your governance program should align your legal, compliance, and tech teams. That's how you lead with responsibility.

**[24:31.112] Bryan Dennstedt:**Love that. And it all ripples out, governance → data → security → workflows → business outcomes. That's what leadership with AI looks like.

**[25:00.821] Bryan Dennstedt:**Tell us how people can connect with you and Global AI Advisors.

**[25:10.100] Sarah Cornett:**You can reach me on LinkedIn. At Global AI Advisors, we help with strategy, governance, implementation, and leadership development. We also offer executive education and speak at conferences and internal events.

**[26:10.112] Bryan Dennstedt:**Awesome. Please reach out to Sarah or me if you need help with your AI journey. Sarah, thank you so much for joining us, this was a powerful episode.

**[26:33.528] Bryan Dennstedt:**This episode of AI with Bry was hosted and produced by me, Bryan Dennstedt. Special thanks to Nancy Velazquez and Kelly Lopez. Video by Leo Moralina. Sound: AI-generated.

🎧 Find us on all platforms at ai.bry.net 📱 Follow @aiwithbry: