Introduction

So I've been to a few MindStone events and George and I were talking about what I can possibly present. I work for a very big company and I've joined the AI team there when we first rolled out AI. And we have a very successful AI program that has been internationally recognized and we're acknowledged in Gartner magazines.

I was thinking, what can I deliver to you guys? And so I've built sort of like a playbook that you can take away of how you can successfully introduce AI and technology in your company.

About the Speaker

So a little about me. My name is Gary Bongiorni.

I'm the head of AI marketing at American Express Global Business Travel.

Please connect with me on LinkedIn. I definitely want to keep in touch with all of you.

Purpose and Philosophy

So before I start, I'd want to be clear that I'm not giving you a strategy for your business. Everyone's business is very unique.

Operational Playbook, Not Strategy

My goal is to share with you the operational playbook we've developed through a lot of trial and error. And then for you guys to take those mental models away into your startup or your business and basically have an unfair advantage when it comes to AI.

Augment, Don’t Replace

So I think it's important to talk about our guiding philosophy when it comes to AI. Our AI strategy rests on one thing and that's augmenting our teams, not replacing them.

So for that, what does that look like? It means taking those boring tasks away from them and allowing the technology to lead and give them their time back. So by reducing the frictions, your employees will have more time to do the stuff that they love.

Why This Matters for Startups

So for a startup, this is even more critical because your people wear multiple hats and time is money. So if you save them 10% of their day with AI, that's 10% more time they get to be like strategic, to work with customers and to focus on growing the business.

Blending Tech and Touch

We call this blending tech and touch and we've got a very effective framework for it.

The Framework Loop

And this is that framework.

So it's a simple loop.

1) Set Rules and Foundations

The first thing is to set the rules and the foundations, a little bit boring, but very necessary.

2) Find Problems, Test, Then Productize or Pivot

The second is to find the problems and then to test your solutions against reality and determine whether to productize your AI or pivot to something else.

So let's break down what this could mean for you.

Laying the Foundation

So when we first launched AI, our AI initiatives, when ChatGPT first came out and generative AI was becoming the thing, we knew we had to set the foundation. So what we did is we created an oversight committee and that was with, we had legal, we had corporate strategy, we had info security, cyber security, HR, procurement, the whole shebang, every corporate stakeholder was invited. And that was to understand how each proposed AI solution could affect the business.

You guys don't have to do that.

A Lightweight Version for Startups

You can do a simple 30 minute meeting that results in a one page document. And the point of that is to write down your red lines about privacy and security before your team goes wild with AI. So it's a foundation for speed and it basically gives your organization the safety net to innovate without fear or breaking something.

Finding Your First AI Win

This is my favorite slide. It's the most fun part to do in an AI program. It's about how to find your first AI win.

Ask the Right Question

So here you shouldn't ask, what can we do with AI? Your question should be, what is the most annoying part of our day?

Crowdsource Pain Points

When it came time for this at our company, we just didn't lock a bunch of engineers in a room and be like, go figure it out. We basically democratized this and we asked every single person in the company to come along for the ride and tasked them with identifying pain points in their life. People really love this actually, because they felt that they were included in something that was very important.

Everyone had heard about AI and they knew that it was going to be important in the company and they could contribute. So it also had a positive effect on them when we did roll out these AI tools and solutions for them, they were actually more receptive to it. And in change management, that's really, really important.

Startup Shortcut to Discovery

So for a startup, this is actually really easier.

What you guys can do is this week, it's a new week, ask your salesperson, your marketer, your operations lead, what are your biggest pain points? What tasks do you hate the most? I guarantee you within the information that they share with you, right there you'll find your first high ROI AI use case.

Validate, Then Decide

So once you've identified your AI pilots and brought them to the team, you've tested them out, it's time to validate the effectiveness of that pilot. So AI is not free and it's certainly not cheap. It has a cost and the more embedded AI is into your company and the tools that you use, the higher cost will be.

Measure Value, Be Nimble

So it's important to ascertain the value of your pilots to determine whether you want to productize or to pivot. So this is where startups have a huge advantage. You're built to be nimble and validating and determining value is very simple.

So you have to basically ask if the costs of the AI outweigh the benefits.

Timeboxing and Measurement Realities

So the key learning for us was to time box. So I do want to talk about this a little bit more and talk about measurement because I'm sure some of you have heard that there was an MIT study that came out where it said that basically it's really hard for organizations to quantify the value that AI brings. So the truth is, it's really hard to measure that. It's very difficult.

Unexpected Sources of ROI

For us, we knew that to ascertain the ROI, when we determined the ROI, it wasn't where we thought it would come from. It actually came from automating all the little administrative tasks that our team hated so they could actually do the real work that they're employed to do.

A Practical ROI Methodology

So I thought, how do we apply this to a startup or a small business? And I contacted some of my friends in Silicon Valley and they gave me this.

So I have to be fully transparent. This is not my or American Express's thinking. This is a methodology where you can identify value from AI.

Metric 1: Hours Back

The first is the hours back metric. So this is an easy thing for you to track and it's the most honest number you can track.

So can someone on your team hold their hand up and say that they got three hours back a week because AI had handled some of the great stuff that Jose had shared with you and your team. That's a tangible undeniable win and that's something that's quantifiable.

Metric 2: Annoyance Removed (Deflection)

The second is the annoyance removed metric, very fun name. Think of this as the deflection rate for boring tasks.

So for us, our chatbot was a huge win, not because it saves some money, but because it deflects thousands of simple questions from our employees to the AI, freeing up the human experts to actually deal with more complex stuff.

So for you guys, this means fewer repetitive emails or fewer manual data entries.

Metric 3: Leverage

And finally, the most important one for people is the leverage metric. So this is the ultimate startup ROI.

So here it's like, if your AI tool or solution that you've introduced save 50% of customer replies to your tickets, that's not an efficiency metric, that's extra runway that you get. So that's postponing a hire and giving you more time to grow.

So that's real value. So it's really important to quantify that.

Where to Find AI Opportunities

So if we go back to problems, which is my favorite part, I love identifying problems because I like creating solutions. Where do you find these opportunities? So there's two recurring patterns that we see when it comes to a goldmine of finding AI solutions.

Pattern 1: Turn Unstructured Data into Insights

So every small business or startup is drowning in a firehose of emails, call notes, meeting transcripts, the whole shebang. And AI is a really great insight extractor. So you can turn all that information and those messy conversations into a centralized source where there's a clear list of actions.

It could be like a list of common customer requests that you haven't identified and that you can now use AI to automate that.

Pattern 2: Automate the Admin Tax

Then every team pays the admin tax, the boring, the repetitive work that has to get done.

And now with AI, especially agentic, this is the single best tool for automating that tax, giving your team its time back, which is the most valuable asset. So if you start with looking at these two things, these are really tangible ways that you can see where AI can come into your business and automate things.

The Human Factor

Okay, that brings me to this most important thing, and that is people. If you remember one thing from my presentation, I really want you to take away this, the hardest part of AI is not the code, it's not the automation, it's not the tools, it's the culture.

The Learning Curve Is Real

It's the learning curve is real for a lot of people. Like at our company, we've got people of all ages, all education backgrounds, and learning this stuff can be quite daunting.

Invest in Education and Enablement

You've got to be patient and really quite deliberate in investing time to educate your workforce when it comes to AI.

We have over 15,000 people and we really create a world first global training program for all of our employees. So they're not hesitant as they go out. So it's very, very important.

Training Beats Tools for ROI

Um, so that investment in your team will deliver a lot more ROI than any tool can be. So that's something I really like you to take away.

Conclusion and Q&A

I know we're almost at time.

So any question answers about this, about AI in general, about how big organizations are looking at it. I'm certainly open to answer anything that you've got.