Subscribe to our newsletter to receive business tips, learn about new funding programmes, join upcoming events, take e-learning courses, and more.
Take the first step to successfully starting and growing your business.
Join for freePosted: Thu 15th Oct 2026
AI is already affecting every business. If you are not using it, your customers, competitors, and suppliers will be.
Its impact is not yet widely understood, and with many businesses adopting it based on a trial-and-error basis, the opportunity to transform your business using a planned approach may be lost. So is your business ready to change? And what can you do to embrace AI?
How AI is being used in business right now
How AI might be used in your business
AI governance, responsibilities, and risks
Dr Paul Corcoran is a leadership, management and innovation expert with over 25 years of experience helping organisations and individuals navigate change and unlock their potential.
He has held senior executive roles as a founder, CEO, and director in FTSE-listed and private businesses in a wide range of sectors.
He is a member of the Institute of Directors Expert Advisory Group on Science, Innovation and Technology, and has a PhD in Artificial Intelligence applications.
In 2019, he founded Interimconsult, where he works with government and industry leaders and provides business advisory and mentoring services to SMEs and start-ups.
Watch more expert webinars
Access a growing collection of expert-led webinars covering marketing, sales, finance, growth and more – ready whenever you are.
Lightly edited for clarity.
Beth Galloway: Hello, everyone, and welcome to today's lunch and learn. My name is Beth, and I'll be your host today. For those of you attending a lunch and learn for the first time, Enterprise Nation is a vibrant community platform for start-ups and small businesses.
I'm very pleased to introduce Paul, who is a business mentor and adviser. In the session, Paul will discuss what your business can do to truly embrace AI. If you have any questions throughout the webinar, please post them in the chat, and we'll do our best to answer them at the end of the session.
As always, this webinar is being recorded, so we'll send the recording along with some follow-up resources later today. Over to you, Paul.
Paul Corcoran: Thanks, Beth. Hello to everybody, and thanks very much for spending your time with me today. I won't go into introductions about myself, as you can check out my LinkedIn profile, but I do want to tell you how I got into AI.
I got into AI in 1989, on the back of an idea I had about doing a PhD to develop an electronic nose. That probably isn't what you're expecting to hear at the outset. It led me to look at two components: the sensors for the electronic nose, and how you process those signals to work out what it is you're smelling.
That second part is the bit of AI that is prevalent now and, to a large degree, hasn't changed. AI does three things: it recognises patterns, it predicts things, and it searches and optimises. And it's all ones and zeros.
Today I'm talking about how AI is already changing businesses all over the world, and the question is how ready you are. What's your state of readiness? There are lots of different types of AI, and some have been around for longer than others.
Things have developed because there is far more data around, computers and processors are much faster, and connectivity between systems is much greater than 20 or 30 years ago. Growth in all of these has been exponential. As a result, there has been a shift from the top-left systems on my slide, such as rule-based AI and machine learning, to the bottom right, where the buzz and hype is now: agentic AI.
I won't go through all the words on the slide, as the deck will be available afterwards, but I want to say something about two of the boxes. Agentic AI is the technology available now that many see as either a facilitator of doomsday or something that unleashes great opportunities and huge benefits for mankind. In essence, it's a set of algorithms that can pursue goals, plan, make decisions and take actions independently.
So it looks autonomous, but it is still based on that digital capability to recognise patterns, predict things, and search and optimise. Down the bottom, I've got artificial general intelligence (AGI), the hypothetical form of AI that might replace us all or be cleverer than us all. In my opinion, and that of many others, it will never be realised, because intelligence is biological and innately human, so AGI is a lot of hype.
It's quite a complex field to understand, but there are some basic things that businesses need to look at. If you look at how small businesses are using AI right now, three out of four are using it for creative content creation. That's the large language model functionality, such as ChatGPT and Copilot.
Just over half are using it for data and analytics, for example, loading up a file of Excel data and asking it to summarise what the data is telling you. Around seven in 10 are using it for admin and support, such as keeping on top of emails, tidying up the inbox and managing the calendar. A smaller but increasing number are looking at automation applications.
These could cover finance processing or correspondence that is more intelligent, functional and personalised than typical mailing applications. The aim is to automate some of those processes and take away the manual labour. What this tells us is that small and medium-sized enterprises (SMEs) are starting with accessible, information-heavy tasks where AI can save time quickly.
For instance, you might ask for an outline for a proposal, or ask it to research a topic and pull out the most salient points. Uptake is higher in larger businesses, and it doesn't suit every business. Not every business needs to use AI.
Smaller businesses tend to use the simpler use cases, whereas large businesses deal with more complex delivery scenarios. Knowledge-intensive sectors such as consultancy, banking and finance have much greater uptake. The vast majority of users are using natural language processing and text generation, which is what ChatGPT does.
It takes language, processes it and creates text, so it knows the rules of grammar and how to summarise and extract information from a long document. Agentic AI is very rarely used at the moment, and that will probably increase. But many of the doomsday applications aren't relevant to SMEs, as they concern mega-corporations, governments and state bodies.
AI is all about use cases, meaning where you can use it. If you talk about the things that might create significant harm, you might then talk about misuse cases, but it's the use cases that matter. The classic one is content generation, which you might use for marketing.
You might also use it for tendering. You can put in the documentation from a tender, ask it to extract the key points, and ask what you need to focus on to get the best score. A lot of that comes down to writing the right prompts, and you might need to pay a bit more for the licence for additional functionality.
If this is your area, it might well be worth the investment, as it's not a lot of money. Then there's administration. If we'd started this meeting with AI on, it would be recorded, and a transcript would interpret what I'm saying, hopefully taking into account my accent and tone of voice.
That is the natural language processing element of AI, and it can be used to create minutes, actions and follow-ups from what we've talked about. HR is another example that often comes up, such as AI interviews that large companies might use for recruitment. Lots of people question whether that is effective, because how you present on screen is quite different from how you present in person.
There are also issues around bias that need to be taken into account when people interact with a visual agent. Increasingly, AI is also being used to automate workflows. If you're a business asking where AI can help, you've got 10 areas there where it might be able to.
It's always a starting point that I talk through with clients. I also talk about three levels of adoption. The first is the assistant level, where people ask AI to summarise a document or do some research.
One thing I always say is to check the references, because there are lots of examples where they don't exist, are wrong, or have simply been created through text generation to add weight and credibility to what comes back. There was a famous case in the United States where somebody presented a lot of case law to support a claim against an airline over an accident on a plane. Despite what was probably a valid claim, he was thrown out of court because the case law had been generated by ChatGPT and many of the references didn't exist.
A justifiable claim failed because somebody misused AI. It's the fastest place to start, but do take care. The second level is the AI embedded in software you probably already use, such as customer relationship management (CRM) systems, accounting, Microsoft 365 and marketing tools.
You can start to play around with that functionality. It could be simple administration tasks that you'd use Copilot for, or analysis of customer engagement through your CRM. The third level is the interesting one, because although it looks more complicated, I think a lot of SMEs should be looking at it.
I do a lot of work in this area. You look at where AI can contribute to your workflows and redesign them, using AI as a tool to improve performance, quality and speed. When you remove a bottleneck with AI, the bottleneck often moves somewhere else, which you need to be conscious of.
Potentially, though, these changes are transformational. Tendering is the classic example that many businesses start with, because it's a heavy, people-intensive activity. The process runs from receiving an opportunity and deciding whether it's for you, through bid/no-bid assessment, pulling out requirements, and identifying where you do and don't have evidence.
You then look at previous material, draft the first response, go through a review, and finish with a compliance check on whether you've answered all the questions. AI can play an active role in that, supporting the information acquisition and processing stages. The key thing is that your business's and your people's subject expertise is unique, and it varies across businesses and sectors.
You still need that expertise, because AI tools give you generalised answers rather than something specific to your question. They are also liable to bias, because what you've put into the tool influences what it generates in future conversations. That can lead to bias and errors in the content.
So you need a subject expert or a panel to look at it and ask whether it makes sense. As I've said, check the references if something is new to you or doesn't sound quite right. AI is a tool, but people need to own the judgements, decisions and approvals that determine whether something fits the bill and will win the business.
Because of the generalisation and bias issues, AI-generated content probably won't differentiate you when you put in a proposal against your competitors. So the gain comes from redesigning the workflow, not just generating a faster first draft or doing some rudimentary analysis of the tender. AI-assisted tendering can add value, but you need to look at the whole workflow, not simply what a tool can do for you.
You also need to pay attention to governance. Depending on how you share information with an AI tool, you need to be careful about confidentiality. There are GDPR regulations in place; personal data shouldn't be shared, and where it is, consent should be given.
If you've got business-confidential information, be very careful, as it shouldn't go anywhere near these systems. The same goes for intellectual property (IP), which you need to make sure doesn't leak through your employees' use of AI tools. You also need to make sure things are accurate.
Copyright is a good example. If you unknowingly reproduce something that is 95% the same as the source information your AI tool used, you are liable for the copyright issue. Using the tool doesn't absolve you of responsibility for making sure that what you put forward doesn't infringe any copyright.
That's the point in the bottom right-hand corner of my slide: the business is accountable for the processes, systems and AI activities it puts together. I do some work for the Institute of Directors (IoD), where I'm chair of the East Midlands region and a member of its expert advisory group (EAG) on AI. We co-authored a publication on what good governance looks like for a business using AI, either directly or indirectly through things such as CRMs.
There are really useful checklists in there, and we can get them circulated afterwards. If you're wondering how to go about this, one thing I always recommend to clients is to look at three high-friction, recurring activities. It's just a good business improvement approach anyway.
Experiment with how AI tools might be used for those activities. You'll find that some things can be done very quickly, but you'll also find that AI often just shoves the bottleneck downstream. Through that experimentation, you get a sense of whether the workflow is better.
You can then ask whether the whole workflow is faster, whether it gives a better-quality outcome, whether it costs less, and what the user experience will be. These things don't take long, and once you've done them, you can look to embed them. One company I spoke to recently set up its own AI team, which wasn't a team of tech experts but three or four people asking how AI might be used in the business.
Their approach was to start small, identify the value, put the controls in place and, when they find something that works, scale it up. The question you should be asking yourself in business is not what AI can do for you, but how you could redesign your system if AI could do some of the work. That's the question I'll leave you with.
Thanks for listening. If you want to connect with me, my details are on Enterprise Nation and on the front of the deck, which will be circulated. I'll now stop sharing and take questions.
Beth: Thanks, Paul, that was great. When you talked about AI sometimes not removing the bottleneck but pushing it further downstream, what did you mean by that? Have you got any examples?
Paul: A good example is the tendering process. 10 or 15 years ago, that would have been a manual review, with a team of people looking at the commercial terms and conditions, the questions that needed answering, and what evidence and use cases you had. To a degree, you can definitely get those pulled together using AI.
But if you look at the workflow and expect that to save effort, what it typically does is put the activity onto a different set of resources. It might move straight away from the commercial team onto the bid writers. If you don't look at the workflow, that part of the business may not be scaled up to generate all the content AI might create, as well as check it against the requirements in the tender.
So you may still need the same number of people dealing with the process, but you've shoved the work further downstream. Subject expert review is another one. Normally the subject expert would write the response, but now the time it takes to review something specialist created by AI might be much less than the time it would take to write it yourself.
I've found myself doing that sometimes. You create something, think you'll rewrite it, and it doesn't take long to start again with a blank piece of paper. It's not necessarily going to save you headcount, but it moves the activity between different types of resources, and bid writing is a great example of where the capacity requirements shift around.
Beth: Absolutely. We have a question from Leo, who asks about your view on using cloud AI or local deployment from the perspective of ease of use, safety and token cost. He also says it was a great session.
Paul: On ease of use, it shouldn't make any difference. What you see is what you get, and where it's hosted shouldn't change that experience. On safety, people talk about sandboxes for AI and some of the big threats you might get.
If you can physically isolate your data and algorithms from the outside world, or make sure the protocol by which you take data into your system doesn't introduce anything untoward, you can make it very safe. It's a development type of question. Nobody can get into that sandbox unless they adhere to the protocol and go through various checks.
On token cost, you're buying a service, so make sure you use it well. The best way of reducing your token cost is to reduce the number of tokens you actually buy or need. You can quite easily waste them through the wrong type of prompts, the wrong type of engagement, or calling on the wrong resources.
Of course, if you're doing big things, there will be a token cost. It's also a bit like a make-or-buy decision, which comes back to the completely isolated system versus using external ones. There might be some things you want to keep in-house and other parts you want to buy, which again reduces token usage and cost.
Beth: That's great, thank you. Paul, we were talking earlier about AI hype. How can a business distinguish between AI opportunities that deliver genuine value and those that are just driven by hype?
Paul: AI is just a tool like anything else, so you need to think about what you want to use it for and whether it will make things better. There are lots of analogies you can draw on. Take an electric drill or screwdriver: it costs a couple of hundred pounds, and if you're doing that sort of work all day, you'll buy it.
If you're only going to hang a picture every now and then, or do something in the shed, you might use a manual screwdriver or an old-fashioned drill. It's the same with AI. It doesn't do anything different to what was possible 30 years ago, other than doing it much faster, in a more connected way and with far more data.
If you don't need things to be a lot faster, and you don't need the connectivity or the huge pools of data, it may not be the right tool for you. So think very much about the use case, and where it makes sense for your business to use AI.
Beth: That's a very powerful point. If someone on this call is starting their AI journey from scratch, what should their first three steps be?
Paul: I would start with understanding where you could use available tools such as large language models, such as ChatGPT or Copilot. That might be for simple things, such as asking it to look at your mailbox and tell you your top three things for today, which is a neat way to get your day off to a flying start. Or you might ask it to summarise a 40-page document that you haven't got time to read.
There is always a degree of checking and validation to do, such as random sampling. But by doing those things, you start to get a sense of the potential in your business. My advice is to play, but play carefully: don't breach confidentiality or do anything inappropriate.
Think about your particular use case and design from there. What does your business need, and where can it improve performance? It's the same question you would ask whether AI was available or not, except that AI is now part of the solution.
It's like spreadsheets and word processors: you found out where they were best used in your business, and now it's a standardised approach. AI will become the same.
Beth: That's very true. We're at time, but I'd love to keep you for one more question to round off the session. Are there any common misconceptions about AI that business leaders need to look beyond?
Paul: There's a lot of doomsday talk about AI. As a business owner running a consultancy, I use it in a way that is better for me and better for my business. A lot of this fear-mongering is part of the hype, and there's a lot of politics around it too.
There are also a lot of businesses going through initial public offerings (IPOs) right now, and overstating the importance of AI will add value to those IPOs. I'll put something in the chat: I've got a paper called AI: The Hype and the Reality, which is a pretty simple read. I'll also put in a link to the Innovation and Change Journal, which you can buy on Amazon and which looks more broadly at innovation and change, as AI is an important part of innovation.
So I wouldn't get too excited about the hype and all the doomsday talk. Look at the use cases you're working on right now that could be changed using AI. And think about the workflow rather than the specific bit you're working on, because the bottlenecks move around in your business.
Beth: That's great, thank you, and a great place to end the session. Thanks so much, Paul, and thank you, everyone, for joining. As I mentioned, we'll be sending the recording and Paul's PowerPoint slides afterwards so you can revisit them later.
Leo says thank you again, Paul; it was a really great session. We'll see you in the next one. Bye.
Paul: Bye. Keep in touch. Any questions, get in touch.
Unlock more on-demand sessions designed to help you sharpen your skills, grow your business and stay one step ahead. Find more Lunch and Learn webinars
