What AI can really do for your business today

0:00This is Britain's growth moment. Do you want to be part of it? On 26th of March, 750 of the UK's most ambitious founders, CEOs and investors will gather in London to shape the future of mid-sized business. The company's driving our nation's success. The Business Leader Summit 2026 isn't just another conference. It's a day built around the three forces that define growth, ambition, policy and capital. And as one of our valued podcast listeners, you can be there in person. To secure a discounted ticket, visit the events section of our Business Leader website and use this code. BLPODCAST, all one word, all uppercase.

0:41We look forward to seeing you there.

0:51This is the theme tune from a major pop culture force in the 1980s. Amid Margaret Thatcher and Ronald Reagan shaping global politics, a show hit TV screens that still has a cult following over 40 years later.

1:29crime fighter played by David Hasselhoff who takes on bad guys aided by his car kit in the iconic show Knight Rider. Even at the time, the show was a continuation of a long-held obsession with robots and artificial intelligence. We've always had a complicated and curious fascination with the subject. Will its rise be a force for good, or something destined to be a sentient being that will be our downfall? Today, AI is advancing at breakneck speed, exemplified in the rise of chipmaker NVIDIA, which has stormed its way into the zeitgeist and taken the crown as the world's most valuable company, all within the space of a few years.

2:06But how can you make the most of AI for your organisation today? How can you measure its success? And can my chat GPT chats and sensitive information appear in Google? That's what we're here to find out.

2:26I'm Josh Dornbrack and this is the Business Leader Podcast. In today's episode, we speak with Matt Weaver, who leads the solutions engineering team at OpenAI. He and his team help businesses of all sizes, universities, governments and other enterprises turn AI tools like ChatGPT into real business results. over 3 million software developers are layering on top of the company's tech through its API and over 5 million people are paying business subscribers

2:56chat gbt sounds like a chatbot i mean the clue is in the name it's not particularly catchy we're not great at naming things and when we first launched it it absolutely was you put some text in and you get a quick reply out now we can't do anything about the name uh because it's such a brand at this point but chat gbt is now so much more than just a chatbot there's a new buzzword around this year about agents or agentic anyone who's kind of been on linkedin or read the news it's hard to avoid what agents but like let's just demystify that a little bit agents basically means that these ai tools have moved from this kind of chat interface where you ask a question and get a response to actually being able to do work for you on your behalf we've basically given the AI access to tools like the ability to search the web or even search your internal knowledge sources.

3:41So, you know, we use Google Drive at OpenAI and Slack. Some people might use SharePoint or Teams to run their business. And by giving a tool like ChatGPT access to kind of reach in and read from those sources, but also to take actions, you can delegate work. And there are a whole bunch of tasks that can be done for you. Importantly, that has to happen safely and securely. A misconception perhaps about ChatGPT, I think most people are familiar with the consumer product, but we have a highly secure enterprise version. It runs on totally separate infrastructure. It has all of the security certifications, data privacy, things that you'd expect, which means that you can trust that your data is going to remain private and secure when you're using it with the tool.

4:21And what does this look like in practice? Weaver gives a few examples. Let's start with Virgin Atlantic, an airline that's really known, their brand is really about their people and about being fun and having this like really personal service um they wanted to use ai to enhance all of the great work that's already done by their by their colleagues and so they've been building into their app a brand new kind of concierge service you know um the the theory behind that was how can we take the experience that customers get when interacting with a member of virgin atlantic staff and extend that into places where you just couldn't be right So within this concierge in the app, you can talk to it using your voice.

5:00It can make recommendations for restaurant bookings near your hotel. It can help you navigate the airport or to find your lost baggage. And so this is something which they've recently announced will be coming later this year. But we've been working very closely with them to really enhance the experience of all of their customers, to bring that signature virgin personal touch to every interaction, even digitally, that their customers have. So that's like an example of building AI into your product or building it into the service that you provide. but there's just so many more examples every single day of where you can use AI to enhance the work that you're doing.

5:32So maybe I can use a personal example for this one because I guess I'm an employee working in a business. Frequently I find myself trying to do planning so we've been going through a planning process for the second half of the year recently and there's spreadsheets all over the place of you know how many people we might hire or what our plans for different regions across the world and expansion might be and I kind of need to synthesize all of this information and basically produce like a simple one pager that I can actually take into the meeting and be coherent. And previously, I might have spent several hours kind of reading through all of this material, trying to make sense of it, writing something myself.

6:08Now I can actually just throw all those files into ChatGPT and ask it to give me that one pager immediately. And usually, I think what's really nice about GPT-5, our newest model, is it'll usually go a bit further than you ask. It will say, I'll just go and search the web and find a few extra references that maybe you didn't ask for, but are a nice touch. And that means that now I just can walk into the meeting with five minutes prep instead of spending two hours doing that, which has totally changed the way that I work every single day. And I kind of can't imagine going back at this point. Now, the thought of putting sensitive company data into an AI tool may be daunting for many.

6:42These concerns have been heightened recently, with news headlines and posts on social media revealing the chat GPT searches have been showing in Google. But OpenAI and Weaver are quick to address this. I'm really glad you brought this topic up because it gives me a chance to slay the myth and perhaps some slightly disingenuous posts. It's true that in the consumer version of ChatGPT, there's a checkbox where you have to click it on opt-in and say, I would like this chat that I'm sharing to be available on Google. If you don't click that box, your chats are entirely private. That option just isn't there for our enterprise accounts and customers, right?

7:17You know, clearly businesses, when they're operating securely and you're putting private sensitive data in that data has to remain private and secure and yours for your own for yours only and so like the data that's put in to any of the enterprise accounts uh with chat gpt that's team or enterprise are not accessible to google and actually we had the feedback we've actually deprecated and removed that feature from the consumer side as well it might have been a useful feature we thought but we took it away because we just don't want people to have that hassle one final thing i'll add on this though is just like super critical is for our enterprise and for our team accounts.

7:51That's the kind of the two business versions of ChatGPT. We don't train on your data. We don't look at your data. It's entirely private. It's yours. It's encrypted. We've got no interest in that. We just want to make the most useful tool possible. And we don't need to see your data to do that or for it to be shared. So it remains completely private. Some studies show that as many as 78 % of UK organisations are using AI tools in some way. However, often this is not used across the whole business, but rather isolated in individual departments. As he works closely with leaders to ensure adoption is embedded throughout their organization, Weaver has developed what he refers to as the golden rule of three, which can apply to businesses of any size.

8:29AI literacy, leading by example, and experimentation. The first one is just around employee AI literacy, right? So I think we'll look back on this time, kind of how we look back on the web, which is like, can anyone imagine doing their job today without access to the internet or without access to a web browser. AI is going to be the new table stakes tool. That's just part of working life. So it's really important that we provide the resources to enable all employees to get up to speed and to enable themselves. We actually just recently launched OpenAI Academy. So anyone can go to academy.openai.com.

9:03And there's a whole series of courses of learning material of videos and webinars, specifically around how to use AI at work, that's now available for everyone. So I'd recommend anyone who's trying to get up to speed on this stuff to go and check out the OpenAI Academy. So a real focus on AI literacy for everyone is key. The second thing, though, around adoption is, I think, around executive sponsorship and kind of leading from the front. Executives and leaders who use AI in public in front of their employees who really champion that and shout loudly about the benefits that they're seeing are going to see amazing adoption across the organization.

9:41I think there is still a bit of anxiety, like, am I allowed to use AI at work? Is it cheating? Will I get found out? I think, you know, those myths need to be busted. And the best way that we've seen that is when execs are just loud and proud about their own personal AI use. And that really sets the tone and drives the expectations for the rest of the organization. The third one is around experimentation. AI is moving so quickly, it can be quite overwhelming, actually. Sometimes you'll hear about all these different models and different companies launching different things. The most important thing, I think, is just to keep trying every day to try and experiment a little bit with some of the newer models when they come out.

10:15One of the great things about GPT-5 is that we've simplified the whole experience massively. I think before now, there were different types of models. You had to know which one to choose for the different jobs you were doing. One of the unique things about GPT-5 is it's just one model. You don't have to think about it and it will figure out for you how to execute the task. So I encourage everyone to go and play with ChatGPT to try new things that you think might not work because you'll never know you might be surprised. and then you'll always kind of be on the front foot rather than kind of assuming that ai is only as good as it was six months ago when things have moved on so quickly so literacy executive sponsorship and experimentation if you do those three things at the same time you'll have like an amazing uplift and the way that employee satisfaction and actually how much people enjoy doing their jobs because you can automate the boring stuff but also productivity and how much you can get done how you can transform the industry that you're working in do you have any kind of pitfalls that you see or come across when it comes to people implementing AI throughout the business?

11:15Yeah, I think equity of access to AI is a key one. Maybe you give it to a handful of people. You might say, hey, I'll just give a license to five people and not everybody else. And then what you don't get is when you do that is this network effect. People over lunch talking about this really cool thing that AI did for them this morning, or I can't believe it. I had this job that used to take me half a day and I just did it in five minutes and when you give AI to everybody and you kind of encourage and incentivize people by executives and business leaders using it in public you get this excitement this flywheel turning of people sharing ideas about how to do things well together so usually we recommend you know don't give it to one person in the team give it to the whole team because they'll actually all increase and be able to share ideas and together one final thing I'll add is a lot of existing software products have like a bolt-on AI feature now.

12:03I think every company is an AI company in some way. And I understand as a business leader, you can get a bit of fatigue. I'm sure everyone's receiving cold calls about some new AI tool that they should be using. And there's an amazing ecosystem of startups out there that have some great products. But I think AI shouldn't be something that you bolt on the side. It needs to be a new way of thinking, a new way of working, a new way of when I see a problem that I'm trying to work on, I think, could I use AI to help me? How might I use it? A little bit, a lot. And then go and go and take that approach rather than it being something that's just bolted onto the side.

12:35So what should you be measuring to ensure the success of AI implementation efforts? Weaver says that the tech's impact is largely rooted in streamlining repetitive tasks, but improvements can cascade into other areas of organization as a result. I think tracking adoption is a good first start. A lot of tools such as our kind of team and enterprise ChatGPT products provide the ability for you to see what percentage of people who have access to these tools are actually logging in and using it every day. How many messages are they sending? Because that's also a good indicator if people find it useful.

13:07It's not just whether they logged in, but it's how much time they spent using it as a useful tool. We actually survey our users, our business users. So across thousands and thousands of employees who've got access to ChatGPT Enterprise, we run these surveys and 95 % of people report saving time every week from using the tools. 77 % talk about how they can complete new tasks or activities that they previously couldn't. And 71 % said that they are more satisfied at work and they just like their job more. But I mean, the metric at the end of the day that I suppose matters is revenue. It's a lagging indicator though.

13:41So in some of those surveys that we ran across AI leaders, we saw one and a half times faster revenue growth amongst those who were fully adopting AI. But that's a metric that we can look at in a couple of quarters time. How do I know if I'm on track right now? How do I measure if this is going well. I think one of the key things we see is trying to survey users formally or informally and just understand how much of the time you spend in your role is a repetitive task. So if this is something you do every day or every week or every month, and that's something that's highly repetitive, we want to try and minimize those things to make sure that everyone's working on the most valuable, the most creative, the most impactful work, which is usually applying our judgment and problem solving and working through new things.

14:25So I think working out how to eliminate those kind of like cookie cutter repetitive processes just by using AI to help is going to be a key part of measuring success. The other way to measure the value that's being created as well is thinking about the volume of whatever matters to your business. So if I run a customer support team, can I handle more customer support tickets with the same number of people? Or can I actually get a higher NPS from my customers because I'm able to give them a much better, more complete service. so I can spend more time with each person. The same is true in finance.

14:55Our own finance team at OpenAI, our CFO, Sarah Fryer, she reported how we drastically reduced our kind of month-end closing process so that her finance team could spend time working on very interesting, kind of more strategic projects rather than that repeatable process. So I think for each team, for each company, the metric that matters to you is going to differ slightly, but the idea is, can you do more? Can you give a better service? Can you spend more time on the interesting tasks rather than the repetitive ones. I think some of the mistakes that you could make around the measurement of success would actually be along the lines of trying to prematurely track or tie together like one specific use case with the outcome of AI.

15:33AI is such a general enabler across so many parts of the job that we do, whether it's writing or data analysis or various different things. So if I say, well, I've got this one use case, I'm going to see if AI can fix just this one thing, that's going to miss a huge piece of the value that you could be capturing really. and it's hard, honestly. All the little things, all the grains of sand add up into a sandcastle when you're talking about productivity. And so it is important to track specific use cases, but don't lose sight of the fact that, you know, employees that we work with report over 60%, save over two hours a week using tools like ChatGPT.

16:09So that's just like all the little stuff that adds up is also super important. What advice would you give to leaders who are really excited about AI, but maybe a bit unsure or about where to begin? What would you kind of say to them? Yeah, I think it's okay and totally understandable to be a little overwhelmed by everything that's happening in AI. It's all happened really quickly and it continues to happen at great pace. I think it's pretty clear now, two and a half years on from the first Chat2BT launch, that AI is here to stay. It's not a new fad or a new craze that's going to die out. And so in order to get started, I think the best thing to do comes back to that experimentation point.

16:46you can read all you want about ai uh in the news you can you can kind of read what other people are doing in linkedin but the thing i advise any colleagues companies i work with my family and friends even is just just try it if you're not sure if it's going to work and i actually found for each individual use users like own ai journey there's usually one moment like a wow moment where they try something which they thought definitely wasn't going to work and it did and you go i I literally heard someone this morning say wow out loud when they tried GPT-5 for the first time because it just did something totally unexpected.

17:21And from that point, people start to think, well, okay, now what else could I use it for? At OpenAI, we call this a feel the AGI moment. Now, what does that mean? AGI is artificial general intelligence. It's the mission of our company is to try and build the most intelligent AI that's as smart as most people would be. and when you feel the AGI that's like you kind of you get a little tingle and a feeling of like wow that is something remarkably intelligent uh that I wasn't expecting uh and it kind of you kind of glimpse the future a little bit of how the world might be down the line so I love it when people get their first feel the AGI moment and then that kind of expands their mind and they get excited about what what could they could do next with the tools you know we've busted one myth shall we say as far as searches being crawled by Google but you know are there any other kind of important compliance or maybe security concerns that you'd like to quash maybe things that people think of that is just not true yeah i think something to be aware of is that this year when this like year of agents in these agentic ais that we hear about they're going to need to have more access into your company systems like the more that you can connect chat gbt up to your google drive or your sharepoint or these different tools the more useful they're going to be i think that can generate some fear around where is my data and how is that being accessed.

18:37I think what's critical, especially around how we've implemented data connectors inside of ChatGPT, is that every individual user's own login is the thing that's used to retrieve the data. So what does that mean? I can't see data returned in ChatGPT. I wouldn't have been able to see if I logged in and went searching for it myself. So we respect all of those access permissions. We encrypt any of the data that gets returned and it remains entirely private and secure in yours. and then for for our slightly larger enterprise customers as well we offer things like data residency so you could choose to have your data stored inside of europe or singapore or india wherever you're wherever your business is operating so i think it's going to be important to kind of like get comfortable with exposing a little bit more of your data into these tools but you can go to trust.openai.com our trust portal and there's so much documentation there about all of the security policies and procedures and we never train on your data people sometimes fear if I put my docs in, am I going to end up with my data being used to train the model?

19:35Absolutely not. For our business customers, the data is yours and it's totally private. No conversation around AI is kind of complete without relaxing some fears that people may have to do with replacing jobs. I know the official line is kind of, you know, it's augmenting jobs, it's helping, but it's definitely a concern that some people have, especially if you're encouraging people in your business to be using AI, you might just create an agent that's just as good as you are. and puts you out of work. But what do you kind of say to those people who may be thinking about exactly that? Yeah, I mean, the nature of work is certainly changing and it's going to keep changing.

20:11And there are lots of parts of our roles which I hope go away. The part of my job that is data entry or the part of my job which is reading long reports to find the small parts that are actually relevant to me. There'll be some types of jobs which change or go away and there'll be brand new jobs that get created. I think if I look back 20, 30 years ago, could you have imagined the idea of an influencer or even a podcaster as a job that existed? So there's certainly going to be so many more things that we can do, especially as these models get good at writing, at developing software and writing code.

20:48I remember I studied computer science at university and I'd always had people coming up to me where they had a great idea for an app. All they needed was someone who could write code to build it for them. and obviously they were coming to me asking for that. Now you could just implement your idea because you can use AI to write the code. And so I think there's so many opportunities. Of course, work's going to change, but that's why it's more important than ever to kind of educate yourself, to experiment, to learn how to use these tools. Matt, thank you so much for taking the time to do this. Do you have any kind of final words for the business lead audience today?

21:22I'll just reiterate the three points that I kind of touched on earlier around how to stay on the front foot and ride this exciting AI wave. Got to make sure that you're educated and that the people around you get access to the material to become great at using AI. Leaders need to lead in their use of AI and share that with those around them and encourage a healthy culture of experimentation. These tools are going to change quite a bit over the next couple of years. They're already really powerful. And the more you're playing, experimenting, the more able you're going to be to take advantage of all the value that's going to be created over the next couple of years.

21:53You've been listening to the Business Leader Podcast with me, Josh Dornbrack. For more business masterclasses and case studies, please check out businessleader.co.uk.

22:10The Business Leader Summit 2026 is set to be bigger, bolder, and more ambitious than ever. Don't just watch Britain's growth story unfold, be part of it. Secure your ticket today by visiting the events section of the Business Leader website and using the code BLPODCAST, all uppercase. It's your chance to join the leaders who are shaping the future of business.