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: Wed 19th Aug 2026
Discover where your organisation truly stands on AI implementation.
Learn how to move from experimentation to the design of trusted, autonomous systems that can scale with confidence in 2026-27.
Know your true implementation stage
Design systems, not experiments
Commit to autonomous scale in 2026-27
A seasoned strategist, architect, and transformation leader with over three decades of experience spanning telecoms, media, technology, and artificial intelligence, John Hauxwell founded Aidentity to deliver trusted data and AI advisory services to organisations seeking to unlock the hidden value within their data assets.
John brings the technical depth of a solutions architect and the commercial clarity of a business leader – bridging the gap between the C-suite and the delivery function with precision and authority.
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.
Ryan: Good afternoon, everyone, and welcome to today's Lunch and Learn. My name is Ryan, and I'll be your host today. For those of you attending Lunch and Learn for the first time, Enterprise Nation is a vibrant community platform for start-ups and small businesses.
Today, I'm really pleased to introduce the brilliant John Hauxwell, who's the founder and CDO of AI Dentity. In this session, John will discuss how to identify where your organisation truly stands on AI implementation, and how to move from experimentation to designing systems that can scale with confidence.
As always, if you have questions, post them in the chat, and we'll do our best to answer them at the end. The webinar is recorded, and a follow-up email will go out later today, so keep an eye out for that. So on that note, I'll hand over to you, John.
John: Great stuff, thank you very much. Hello, everybody, nice to meet you all. I'll just switch on my slideshow so you can see what's going on.
I hope you enjoy the presentation and conversation. If you have any questions, please wait until I finish the presentation, or pop them in the chat, and we'll discuss them at the end. Thanks very much – so let's begin.
AI at the moment is very much a strategic discussion within most businesses. A lot of it is: is it hype, is it a bubble, is it real? Does it help, does it not, are we doing it properly or are we not?
These are lots of questions that we unfortunately don't know the answer to. This presentation will give you some ideas around how we answer those, and what you need to do to put things into perspective within your own business – so you can move from simply running pilots to a fully scalable execution within your business.
So, the great AI reckoning – a very simple thing. We are in the age where everybody needs to understand what's going on, so we need to end the experimentation.
We need to communicate what AI is around our business, and have proper communication about it right the way down from the lowest levels to the CEO, and from the CEO down to the lowest levels. It's no good just going one way and expecting it to work, or trying to come the other way, because then you don't have the right levels of approval necessary to make things happen and implement them.
So if you're the CEO and you've got maybe two people – but if you're a slightly bigger SME, let's say up to 100 people – you really have to think about communication. You have to think about stopping working in a little siloed environment.
Start thinking about how you communicate across your business, what the data is that you use for your AI across your enterprise, and how you integrate that data and make sure it all reckons correctly, so you have a single point of accurate truth.
This is important, because nine times out of 10, people are running shadow AI or side information, and they end up with different truths. They may not be, but you should have a single point of truth that's qualified throughout the business.
You need to end people randomly running prompts and move to rigorous system design. Make sure everybody understands the criteria, make sure you write your use cases, and make sure you understand the technical requirements and the outcomes, so you can define what you're doing and encourage AI competency and literacy within the business.
As you can see on the last point here, 80% of pilots fail to scale. Why? Because of a lack of data and structure, misaligned incentives, and absent governance – I can't stress that enough.
Make sure your data and your process for AI is well governed. These are the silent killers of AI ambition – you might as well put your AI in a room full of carbon monoxide and watch it die. If you don't do these things, you will end up in trouble.
So if we talk about the new standards here – from chat to autonomous workflows – the interface is about agentic AI. It's not just a product; it's about end-to-end operational control without human interaction or prompting, but you start to have humans in the loop to ensure quality, success and accuracy.
So it's no good saying we're going to cut humans out entirely – you can't do that. But you can start to really incorporate AI into your workflow and manage baseline tasks based on expectations, so you become competitive.
This can really start to allow people who were doing mundane, process-oriented tasks to actually start doing something more creative, and think about ways to really improve your business, rather than being tied down in the mundane.
This is very much the story of Salaria, as posed by Isaac Asimov back in the 1960s – the idea that if humans were freed of mundane tasks by robots or AI, they could then begin to develop more creative and insightful careers, rather than just being tied down to a desk.
So, we need to think about AI and profit and loss – you need to think about the capability and your business outcomes. A pilot should be tied to successful business outcomes.
If they're not measured, and it's not correctly defined for its impact, the AI will fail. If you don't know what you're measuring, and you can't measure it, there's no way to check your results.
So it's very important that you know what your KPIs are, and your KSFs – key success factors, sorry – for the launch of it. If you're going to include that, then you want to include it in any OKRs or business objectives as well. This is very important.
So here we are now in 2027, almost at the end of it – where has this year gone? Let's think about the reality check: storage versus content.
The context of it – how does it work? If it's just stored randomly and dumped in a data warehouse somewhere, or on a flash drive under a desk, then you don't know what you're looking for.
You don't know what the data is actually showing you, and it's not being properly governed and managed. So it's important that you understand the data's context and its construct, and how it works, so your AI can actively use this data and get the best from it.
So you need to fix the context window – moving from unpredictable, prompt-dependent outputs to reliable, engineered context pipelines. It's very important that you think about how your data is structured, how you want it to be utilised, what outcomes you expect, what outcomes are delivered, the delta between the two, and what this means for your business.
It's very important that you think about this context challenge, because then you have to think about how your data is structured and what the technical challenges are for that. Is it a data warehouse? Is it a data lake? Is it perhaps a data mart?
So you've got lots of different data silos, with people governing themselves and bringing it all together using some form of lambda-type query. It's very important that you think about these things and understand what that technical context means for your business, in terms of people, cost, expenditure and time.
How quickly this can be performed allows you to integrate AI successfully and rapidly, with the right results being profitable for your business. So, as I said, the human role shifts from doing to deciding.
You set the intent, review expectations, and govern the outcomes, rather than executing the tasks. Task execution becomes part of the AI.
So if you're running a structure of AI agents, how that data is passed between them, and how that task gets executed, is not your responsibility. That's the responsibility of the AI and the responsibility of your data engineers.
Blueprinting – the autonomous blueprint. We have to think about adaptive operations, digital twinning, and predictive capital deployment.
These are all things that need to happen inside your business, particularly if you're in a manufacturing or engineering space. DevOps – leveraging self-optimising workflows to reduce friction, making things frictionless within your business – compressing the cycle time, reducing time to deployment and delivery, and driving sustained margin expansion, i.e. profitability.
How do you make sure that your AI really works across your business, compresses cycle times – i.e. reduces the time from initiation to deployment – and how does it improve your profitability? These are things you need to look at, measure and understand.
You need to understand them thoroughly before really getting into developing and deploying your AI. These are things you need to consider before you start spending money.
Digital twins – does everybody know what a digital twin is? A digital twin is basically an engineering-created blueprint, or model, of what you're doing in the real world.
This could be a jet engine or a wind turbine. Every single measurement that a sensor has, coming off that device, is deployed and shown in your digital twin.
So if something is going wrong, you don't have to climb up a 100-foot mast to inspect a blade – you can see it inside your digital twin, inside a virtual environment. This speeds up deployment fixes, reduces downtime, and is really, really good.
The simulations beyond that are becoming very effective too. There are a lot of companies that people use for other purposes, like AutoCAD, for example, that have been heavily involved in developing models for digital twins, incorporating sensors, and feeding that back through communication channels into the model, so people can see what's happening and if there's wear and tear.
And if you're looking at real-time capital – real-time risk agents monitoring markets and moving capital around – that's basically web trading. This is predictive modelling; it's been around for a while now, but it's becoming quite prevalent.
It's about time that we start to see the results inside the normal world of people, rather than just locked away in the City. We're starting to see that now through AI – it's starting to happen, and it's becoming very, very effective.
So, the European landscape – there's a lot going on. The EU AI compliance act – I call it a compliance mode, because it sits around your protected fortress, and it's there.
Unfortunately, it restricts you going out as well as coming in – it's difficult. Compliance for that becomes a global differentiator.
You need to understand that act, and understand how your business fits within it, and that you are compliant and that you do understand it – because if you're not, it can get very, very expensive very, very quickly.
Let's talk about the talent gap. In the UK, we have some great people, but unfortunately, not all of them are here.
So if you're looking to upscale or recruit, Germany, Austria, Switzerland, Benelux and the Nordics have far more available people to recruit or employ than we do in the UK. Unfortunately, they are also then more expensive.
But if you're looking to do this quickly, it may be better to hire externally rather than recruit internally – just hire someone on a short-term, part-time basis as a contractor and bring them in to do it, rather than trying to recruit them full-time. Do you need a data governance service full-time? Probably not – you might be able to bring one on an interim basis.
And then let's talk about how you think about making decisions from your data. Your data has to be accurate, and the intelligence that it provides should help you to provide real-time strategic actions, so you can really start to think about what you're doing and make an impact in your business from your AI.
Most people think AI will tell them to do the mundane, but if you use it in a much more business-orientated manner, you can really use it to make some great decisions and some great money. But again, that's based on accurate data – if your data is inaccurate, you're going to spend weeks trying to get it accurate, running around asking your technical people to redo a load of data, ETL or ELT for you. It's terrible.
As there's a few comments on the right here – EU regulation rigour is obviously a burden. However, if you architect for compliance, you will outmanoeuvre those that try to retrofit it.
Same with data governance – don't try to retrofit your data governance. Start early and run it from the beginning; it's much easier than retrofitting data governance, AI governance, or compliance.
It's like trying to untangle someone's cooked spaghetti – it's not the easiest thing in the world to do. So it's very important to get that done from the very beginning.
So, AI implementation – most of us in our AI deployment are at stages one to three, at the foundation: just understanding the basics, automation of systems, and experimentation of ideas.
However, unless those are correctly reported and accurately analysed, they're a waste of time. If you do that properly, then you can start to look at connecting systems and using a pilot deployment to see how well that goes – this is integration, so it's stages four to five, top left.
You can then keep it simple, run a few pilots again, test the results, and check that they meet what you expect. Check that these fit with your business objectives, and that they could fit with different departments – marketing, finance, production, etc. – and avoid the complexity trap.
Don't start to think, 'oh, I can deploy this everywhere now', because you can't. You've probably got a lot of legacy systems, with legacy data architectures and solutions that are being run in an Excel spreadsheet under somebody's desk.
If you start to have what we refer to as technical debt – where you've invested a lot of money in systems that are now becoming redundant – the process stalls, because you're worried about spending far more money than the results you're achieving. So you need to start looking for the correct processes that maximise your yield, as we talked about earlier – financial responsibility.
And then the autonomous scaling – so agentic workflows with profit and loss accountability. Your process becomes automated, becomes autonomous, and is self-thinking and self-regulating, and manages the profit and the capability of the business.
If it's creating a loss, it will cut that programme and look for ones that are going to build a profitable workflow for you. This is the advantage of AI.
So, a quick execution road map of where we are now. Now, to the start of next year, or Q1 next year: audit the infrastructure, identify your workflows, identify your requirements, and consolidate fragmented data into unified intelligence layers.
That's scary, because most people just think, 'oh, I've got a database over here, another database over there, a spreadsheet over here' – they don't talk to each other. Maybe you think you'll just link those together with Copilot, but you've got to think about what that data represents, where it's coming from, whether it's trusted, accurate and complete.
All the things you think about for your data governance need to apply to any data you're going to feed into your AI. And then your AI needs to think about it in the same way – is my workflow complete, is it accurate, are all my sources accurate, are they complete, do I trust it, do I trust them?
Once you've gone through all of that, and you've got it all sorted out, then you can start to think about scaling into profit and loss targets, and transitioning teams from AI-curious to AI-productive through structured enablement – training, management, coaching, mentoring.
All these things need to happen if you're going to start to make your business AI compliant. People need to understand what's going on, and just because you deploy an AI agent doesn't mean it's going to get used, or that it's going to get developed or deployed correctly.
So you need to have a very realistic view of what's going on – and if it's your baby, particularly if it's your business, you want to be very much hands-on with this. Keep tight control of the outcomes, the inputs and the outputs, and make sure that the outcomes fit with your business direction, your mission and your vision for your business.
So that takes care of 2027. 2028 looks like full workflow integration – are you being competitive? Are you beating your competition because of AI deployment? Is it making you more financially viable?
Build what scales, not what shines – I'll stress that again. Just because it's shiny and looks lovely doesn't mean it's going to be any good to your business, and it doesn't mean it's going to be accurate in how it works through your business.
Build what really matters and what really scales, rather than what looks pretty and might be something to shout about once it's deployed but doesn't really bring any benefit. Again, you need to think about how these are weighted in your KPIs, and within your OKRs and key success factors for your business.
So – multi-agents. Everyone thinks, 'yes, I've got an AI running, I'm running Copilot, that's gonna be fine' – but just because you're running one doesn't mean it's going to give you the best results.
Just as you have multiple teams – a finance team, an operations team, a product team – all working together to produce the best results, it's the same with agents. Each agent, as I say, is defined by a different role.
Claude strengthens the argument, ChatGPT fact-checks, and Gemini refines the output. The human in the loop is the conductor, setting the tempo, resolving conflicts, and ensuring the final output serves a business objective – then you can leverage this at scale.
If you start trying to think about this as just 'let's deploy it and go', you're not going to get the results you want – it's going to be a little difficult. It's much better to check at each stage that you're getting what you want, rather than run large costs, a large deployment, and lots of time and effort, and find out it wasn't any good – when that could have all been determined at stage one, rather than deploying to the final stage and finding it doesn't work.
So, obviously: check once, check often; test once, test often; fail often, rather than fail once with a large cost.
Noise – AI is a very noisy place. What's driving real ROI? Workflow operations and automation, predictive analytics, customer intelligence, risk compliance, and supply chain automation.
These are key things to think about, and they will drive real return on investment across your business. So every AI investment must be benchmarked against a measurable business outcome.
Very important – what are you measuring? Can it be seen? Can it really be measured? Does it make sense?
And then there's a little piece in yellow – stop making the same 10 mistakes: unclear ownership, absent baselines, over-engineering, zero governance, and people saying 'not my responsibility, it's not my job'.
Make sure that's all defined and laid out carefully before you even get into deploying AI. Make sure everybody understands their roles and their responsibilities – there should be a RACI for every single one.
And within the EU AI compliance act, it states that a data owner must be there for every single AI process, and must be operationally responsible for the outcomes of that AI. So you have to be there and be able to stand up and say, 'this is accurate, and this is right'.
If you're unsure, ask somebody to come and help you, and get them to work with you to ensure that you can sign that off with clarity, and be responsible, rather than having to fudge the details.
There is a massive talent and culture shift as soon as you start to talk about AI. We talked about closing the skill gap – invest in structured AI literacy programmes across all levels, so everybody knows what's going on.
Strategic fluency is now a leadership requirement, so make sure you know what you're talking about so you can communicate it to everybody else. A little knowledge is a dangerous thing, so don't do that.
Make sure you can really talk with clarity and confidence about the process and how it works, and give an accurate and coherent account, so that everybody understands what's going on.
Governance and transparency – lead with clear AI governance frameworks. I've actually built a few of those; they're complex.
And if you're a small business, maybe they're not as necessary as you'd think – but you have to define it and walk the walk. Do you need to have all the documents in place? Yes, you do, because it's easier to start small and not have to retrofit.
Because if you don't do it, you're going to have to do it later on, and it might be more work and more complex to put it in place after the event, rather than doing it at the beginning. So don't think that it comes later – think now, and think transparency.
Transparency builds trust with employees, regulators and customers, and that's exactly where you want to be. You can't buy trust – trust is something that really defines your business, yourself, and your relationship to everything that you do.
Without trust, you have no transactions; without transactions, you don't have any sales, and you don't have any value. So trust is vital, and AI is a strategic asset.
View AI as leverage, not a threat. Think of culture and adoption as a multiplier that separates good implementations from great ones.
If you really get culture and adoption right early, and develop it so that people feel comfortable and trust AI – and trust you for deploying it, and don't feel threatened or feel their jobs are at risk, but instead feel enabled to do different things within their work – they'll start to think about the business in a different way, and bring value rather than just doing repetitive tasks.
It really creates a strong ethic that binds the business together, and allows trust to become not just something you espouse, but something you actually deliver.
Financial impact – working capital efficiency. We all know how important working capital is for a small business.
So let's think about how we use it to optimise our cash cycle: payables, inventory, ordering, process management, billing, etc. Then we start to think in terms of risk mitigation and margin protection – exposure, anomalies, and protecting our positioning, making sure that we have the right margin.
And then, long-term value for shareholders and owners – so we can align them with growing sustainable capital, sustainable returns on our investment, and sustainable business growth.
So these are the key points – if we look on the right there, from cost to value creation: AI helps us create value, but it also helps us capture value from within our business and from within our data. So we look around and see what value we can actually capture, not just what we can create.
But I focus on value creation here, because eight to 12% of revenue is a lot of money for most start-ups. When you put that into perspective, it's a lot of working capital, and it's good to have.
So protecting our business and thinking about how AI can help us create and capture value is really important.
Strategic infrastructure – what does that mean to us? It's not a technical decision; it's a strategic one.
Organisations that operate should architect their data as an intelligent asset, build workflows to scale from the very beginning, and drive that through the entire process. You need to start thinking about where your data is, how it's managed, how it looks, and whether it's coherent.
So a good data and AI governance framework has to happen right from the start. We then think about the workflow engines – how does that data get into our AI? Is it scalable? Is it robust?
Are the agents performing correctly and producing the right outcomes for us in their process pipeline of operations? And then we check for regulatory adherence, and embed that at every layer.
We have to be compliant – it's not just a case of being 'regulatory near enough'. No, you have to be; it's vital.
Organisations that aren't will be shut down and closed, and won't be able to trade, or could be fined and end up paying a lot of money in fines. Compliance is not an option any more – it's required.
So start thinking about that from day one, rather than thinking that compliance can be bolted on later or retrofitted – it can't, not really. You need to think about it early, and adopt the right processes and controls.
Competitive advantages – AI goes from a cost centre to a growth driver. To deploy AI initially can be expensive, particularly if you're doing all the governance, the data management, the data accuracy, your data governance, your AI governance, and your measuring, controlling and monitoring.
It's a lot of work, and if you need to get technical people in to help you deploy that, maybe you have to hire – so it's a cost centre initially. How quickly does it become a tool to generate revenue and margin expansion, and show clear profit and loss outcomes?
The market is crowded with AI adopters, so disciplined systems thinking – not just tool access – and a deep, detailed process helps you to create defensible competitive positions.
As I said here, the decisions made today on data governance and workflow architecture determine the ceiling of what your operation can achieve in 24 months. So you really need to start thinking about that today, not tomorrow, and start acting on it, so that further down the line you're in a stronger position than your competitors.
So, a quick summary of this: the era of experimentation is dead. Pilots without profit and loss accountability are a luxury in 2026, and no one can afford them any longer.
Start making them so they're not just experimentation, but pilots that deliver real value, and capture and create real profit and loss within your business. Then we think about point one: autonomy is growth.
Workflows are the primary levers for scalable and compounding business performance. If you start here with a little AI, and move it all the way along through other AI agents, the workflow looks good, the profitability racks up, and the costs decrease, because once you start to do one process, you can reutilise some of the tooling.
So you start to get real business performance, and it's scalable, because you're literally able to reutilise a lot of the work you've already done. And once your data governance is done at the bottom, it's done for everything – so it's a good thing.
And one differentiator remains: outcome-focused execution, strongly disciplined and well managed. That's the only thing that matters – everything else is just noise.
And people say, 'oh, we don't know this, we don't know that' – no, get them to start thinking. Get everyone to focus on the task at hand and be really ruthless with it.
Like all good pilots, unless you're already focused on the outcome, it will just wither and die. So be strong, be focused, and don't allow yourself to deviate – as they say in Star Wars, stay on target.
So, what next? Understand really where you are – assess your organisation and where it truly sits across the 10 implementation stages we discussed earlier, not where you aspire to be.
Are you really still in discovery and foundational stages, or are you actually in deployment? Start thinking about that honestly, rather than aspirationally.
Identify one workflow – one single, high-margin, agent-ready workflow to prioritise. This could be an ordering process, something to do with your CRM, or your communications and marketing.
These are all things that can be delivered quite quickly and quite well by AI, even in its still fairly infantile state, where we are now. Narrow your focus and be really objective – stay on target.
Drive for better results rather than broad experimentation, and set yourself a time frame – 90 days, 180 days. Ninety days is good, because as I said, fail fast and fail often, rather than fail once and fail expensively.
Commit to measurable financial outcomes for your first project – no target, no accountability; no accountability, no scale. Also make sure that people are responsible and accountable for the task, consulted and informed, so you've got all the information at hand and readiness.
Make sure everything's signed off and approved, and that you've measured it, quantified it, and are able to deliver it. Make sure it works, make sure your use cases are correct, and make sure everything's captured – that you do a proper project initiation document for each workflow.
Focus on what scales, and not what shines. The most impressive AI demos rarely translate into production value.
Prioritise workflows with compounding returns over those with the strongest conference narrative – don't just tell a story, show value. Don't just say 'this is pretty, this is nice' – it may be ugly, but it brings real value.
Show the value, don't show the prettiness. And when you're doing things for yourself, show the value to yourself, not something that looks lovely in a pretty PowerPoint.
Stop chasing the tools – build a system. The tools are all there; everybody's talking about new Claude engines and everything else, all the MCP-type stuff – it's a minefield.
Stop listening to all the noise about tools. Think about what it is you want to achieve, and build a systemic view, and then deploy that – rather than thinking, 'oh, this tool's pretty, this one's nice, this one's cheaper, this one's expensive.' Stop doing that.
Think about it from a very logical and pragmatic view – be really aware of what it is you're trying to achieve, rather than what looks pretty to your fellow founders, shareholders and management team.
And commit to the transition – think about it: 'I've designed this, I'm going to use it, I'm going to deploy it throughout my business, I'm going to encourage the business, and I'm going to work with people to deploy this accurately and utilise it to get the best results.'
From consumers of AI to designers of AI systems – we're all there, and we can all work with that quite happily. This is the transformation that separates leaders from laggards – don't be a laggard, don't be that final percentage.
We think about early adopters, main adoption, and laggards – we all know that curve. Don't be in the latter part of that curve; be mainstream, and focus on what other people have learned.
Learn from your own deployments and your own mistakes, and really bring it together. Start to listen and talk to people as soon as you can, particularly those involved in AI and AI governance, and data governance, about how this can work for your business.
The difference between winners and losers will be the ability to automate the execution of strategy, not just the generation of content. So rethink what your strategy is, how you're going to deploy it, what those measurements are, and what they actually say about your business.
Don't just generate pretty pictures, and don't just sit there and draw a PowerPoint – actually think about it. Think about your business, and where the advantages are for you against your competition, and where the disadvantages are for them.
This is the inflection point. Organisations that act with architectural intent today will own the market next year.
Start thinking about what it is you want to do – get those blueprints down. Speak with your systems people, speak with your data people, and really begin to construct what it is you want to do.
Talk to people, find people to talk to who can help you with this, and reach out. So – are you ready?
Execute, then: audit your true implementation stage, design your systems rather than just experiments, and commit to autonomous scale in 2026, to be complete by 2028, and thoroughly throughout your business by 2030.
Are you ready to lead? I hope you are. So that completes my presentation – I hope that was useful to you.
I'll take a quick look in the chat and see if there's anything in there.
Beth: Hi, John.
John: Hey, how are you doing?
Beth: Hello, yeah, all good, thank you. So you might have seen a change of face – Ryan had to leave, so I'm here just for the Q&A, if that's alright, John?
John: Sure, great.
Beth: So yes – if we could have a few minutes for some questions, that would be amazing. So, John, how can organisations objectively assess their current AI implementation stage?
John: Wow, okay. I would start with: how much do I know? Do I know what I'm doing, or am I just following a trend? That's where I would start.
I would say, how confident am I in what I'm doing? Use that on a scale of, let's say, one to five, like a maturity index – zero, where I have absolutely no idea, and five, where I know exactly what I'm doing.
And be honest – anyone who gives themselves a five should be working at Google Mind. So, no: most people overestimate their own capability. Start with that.
Ask yourself that very simple question, and ask your teams: are you confident with the AI we're developing? Do you know what it's doing, do you know what it's for, how are you working with it, are you utilising it? These are simple questions.
Then start looking in a bit more detail about your data – where is it, how is it formed, how is it stored? Is it literally on an Excel spreadsheet, under somebody's desk, on a laptop we've had for about 10 years? That sort of thing.
Start really asking the questions, and ask your teams how they're organising their data – again, one to five. This is basically a maturity model; I'm sure most people are familiar with maturity modelling, but if you're not, please talk to me, and I'll help you.
I might bill you for it, but I'll help you. Maturity modelling is important – you've got to understand your data, understand what people are doing with it, and understand what you're trying to solve for.
What are you trying to do? What are you trying to achieve? How are you measuring that – all the things we talked about today.
If you can't measure something, you don't know what the results are. No measurement, no use. If you don't know that something is driving your profit by 5%, why are you doing it?
If you don't know that something's dropping your cost by 3%, that your goods are turning out 24 hours in advance rather than being late, and therefore increasing your productivity – these are all things you should know and should be monitoring.
So you need to really think about quite a lot of things to properly assess where you are. But I think the initial discussions with your teams and your staff, understanding what you're trying to achieve – if you do that first, and are honest with yourself, that'll give you a good idea of where you are.
Beth: Yeah, that's great. So you mentioned AI maturity there – so what, in your opinion, is the biggest, or most common, misconception organisations have about their own level of AI maturity?
John: The most common one is: 'yes, we're using AI, everybody's using AI.' The fact is that they're a Microsoft house, for example, so everyone's got access to Copilot.
The only problem is everyone's running Copilot independently – they're not running it on a single source of data, or as a single source of truth. They're just randomly pulling information from the internet, churning it all together, and shoving it out into the business environment.
And while they're doing that, they're also putting company-sensitive information into Copilot and shoving that out into the world – whether they realise it or not, it's a two-way street. So company confidentiality goes out the window, business integrity goes out the window, and trust goes out the window.
This is some of the biggest problems – but the fact that people are running what we refer to as shadow AI, rather than managed and orchestrated AI, is the biggest problem and the biggest threat.
And once you start getting into a situation where finance has one number, marketing has another, and the CEO has another – and it's all supposed to be the same thing, like 'how many sales did we do last month?' – and everyone's arguing about that, you know you have a problem. By that point, it's too late.
Beth: Absolutely, thank you, John, brilliant. Yeah, I think that's all we have time for today, unfortunately, but I think there's still so much more to cover.
I'm sure Ryan has popped your Enterprise Nation profile, as well as your LinkedIn, in the chat, so everyone on the call, please feel free to reach out to John – I'm sure he'd be happy to answer more questions.
Yes, we just got some comments in saying 'brilliant session, thanks so much.' Yeah, that was great, thank you, John.
And just a reminder to everyone on the call today: we do send out the recording with further resources later today, so you will receive the email after that. As I mentioned, please go ahead and connect with John.
So, thank you so much for your time, and we hope to see you in the next one. Thank you – thanks, everyone, for joining. Bye.
John: Thank you all, bye-bye.
Beth: Bye-bye.
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
