Practical considerations for AI adoption: Part 3

Miriam: So Ash, what we’re seeing with a lot of our clients is that they’re really keen on diving into AI. Let’s leave aside the fact that sometimes they don’t know what they really mean, but they are a bit wary of jumping into a major project without fully understanding it. They have seen the presentations from all the very big technology consultancies, and they know they want to do something, but they’re not quite sure. What are the kinds of questions they should ask? What should they be looking for if they want to get started?

Ash: Good question. I think as well – just to pick up on that – as an organisation we do actually do quite well off the back of some of those large organisations that got in and created the slide deck and then walked away and then told them how much it’s going to cost to implement it, and then they may well take a look at some point. But I think the questions need to ask is, is what am I going to be left with? What am I going to end up with when that organisation rides off into the sunset? Or almost the binary opposite of that is; how do I get this organisation out of my organisation? Because I’ve become so beholden to them.

I think one part of that we have built over the years at Ten10 is this idea of almost making ourselves redundant in this process. You know, we don’t want to stay there forever, and in actual fact our academy was built with that in mind essentially. To answer that question of; “Right, you’ve got these people, but what happens when they ride off into the sunset?” To which the answer is: These junior individuals get to transition to be your permanent employees.

So these are individuals trained in the skills you need, trained in AI, who are the right fit for your organisation culturally, from a personality point of view, in terms of alignment of values and goals. And then over time, the seniors do roll off and you’re left with those juniors. One of the challenges is always how do you build that capability if you don’t have it already? So within Ten10, we have those senior consultants who are able to come in and define that and then build that and put the meat on the bones. But then over time they do roll off at a point everybody’s comfortable with and happy with. Leaving those junior people behind to kind of take up the slack.

​​Miriam: Having been on the business side, having been the CFO who signs off the consultancy checks, sometimes over many years. I absolutely love that model because it just solves so many problems in one go.

It solves the problem that the organisation has trying to hire the right kind of people, then to develop them without having the skills internally to develop them – you provide all of that. And it solves the problem with having skills in-house that can continuously grow and evolve and learn, rather than each time a new change comes in. You have to go outside and bring the consultancies in, because having seen many business transformations, that have to be redone every three years and all the consultancies come back – yet again – with a lot of people who then have to be educated on the culture of the organisation, on the structure, on how things really work, etc., etc.. It’s a very inefficient process. So I love that model. And as I said earlier, it’s very aligned to the way we think about our work, in terms of the aspect of “change management”. I’m not even keen on that word, because I don’t really want to “manage change”. I want to help people in the organisation develop the capabilities to manage change themselves, to continuously grow, to be a truly learning organisation.

What do you think are the very practical first questions that somebody should ask to get started with either the technical side or the people side?

Ash: I think one of the first things is, like we discussed last time in terms of helping to define that pilot. Now, the way in which we do that is, we just come in and there’s kind of a fixed price for a piece of work to be able to evaluate what you are looking for, to be able to kind of put some boundaries around it, you know? Exactly as you said before, Miriam, in your sort of finance space, being able to constrain that and say, this is what it’s going to look like, this is what the outcomes are going to be. But then also being able to have ultimately, after that pilot is being conducted and after the value has been demonstrated, to then be able to have more junior, lower-cost people come in and be able to do that.

Miriam: Oh that makes my CFO heart sing!

Ash: Yeah I hear a few cheers from our CFO as well whenever I mention that. But one thing that I think is really important that you touched on, that was the idea of a learning organisation and continuous learning. One thing that’s really clear to me is that AI is changing things at such a pace, that I think the days when an individual says, even if they’re not a technical individual, if you’re a business analyst or a project manager says; “I know my stuff and that’s it. Now, I don’t need to learn anything new”. I think those days are gone. One thing we see when we land junior people who we test in the system for that continuous learning mindset, and instill that into them, is that they then take that into the wider organisation.

So we work with some organisations where the team’s average age is quite old. You know, they don’t have an attrition problem. Nobody ever leaves. But that creates its own problem in terms of the aging workforce, people often quite set in their ways. But when they have these junior people working with them, with that kind of mindset that gets adapted by the wider teams as well.

Then people realise that there is a way to do this, and as you said, it’s non-threatening. It’s not a case of; “Oh, gosh! My role is changing. I must learn this”. I think I would argue that’s one of the most important things.

Miriam: I would absolutely agree. I personally dislike the phrase; ‘people don’t like change’. Because the reality is, people do like change. I mean, people go on holiday every year completely changing the daily routine. The vast majority of people do like change. But they don’t like to be changed, which is also why I don’t like the term ‘change management’. When you create that organic dynamic where you tap into the motivation for people to change. And by the way, that is not just one motivation, there’s obviously a whole plethora, but you can tap into the individual motivations. People actually quite enjoy the change. I’ve seen this in organisations where senior people or more longstanding members of teams were very set in their ways. I have seen that when some more junior people come along with new ideas and it’s integrated in a sympathetic way, they take great pride in both teaching, but also learning. They actually really enjoy that. That is the vast majority. So this image that 80% of an organisation’s workforce doesn’t want to change; I think that’s completely wrong. That’s a picture that’s painted because of traditional approaches to ‘change management’, which I think are very much on their way out, because with AI they don’t work very well.

Ash: No, I mean who would welcome having somebody come into an organisation and tell them what they’re doing wrong and how it needs to be checked? You know, it’s a classic sort of way of implementing it. But that still is very much the case. Rather than doing it internally and building it slowly.

Miriam: If you think about it, it’s a very old fashioned way. It always makes me think back on the 1950’s and 60’s, where knowledge was very hard to acquire. At that time a consultant who understood a lot of things about an industry could take that knowledge into different organisations, because they themselves didn’t have the ability to acquire that knowledge.

Well, these days it’s gone! Give me 20 minutes and I’ll have the whole McKinsey playbook in front of me. So it really emphasises that these days the implementation which relies on the human interactions and the relationships, and also human ingenuity becomes much more important. Which you can’t demand from the outside. You can’t even demand it from the top down. It needs to be a much more organic process.

Ash: We’re at a place now where businesses are starting to adopt it. They’re running pilots. They’re facing some challenges with that, which we’ve addressed in terms of how you scale it from a people point of view, as well as a technical point of view. But what sort of changes do you foresee coming down the line in terms of trends?

Miriam: I’m not going to comment on any technology trends, because with AI the change is so fast that whatever I say now will be outdated in probably two hours’ time! But the big change that I think will accelerate is the democratisation of knowledge – and to an extent – skill as well. What I mean by that is that with the use of both personal productivity AI as well as Agentic AI, more junior people in organisations will start to take on certain tasks that were traditionally reserved for more senior or more experienced people.

Obviously, this has to be done with the right guardrails and the right kind of support. But we’re already seeing in some organisations that, for example; junior lawyers, now doing work around client acquisition that was previously reserved for more senior partners. They’re able to do that because they are able to learn about the industry, learn about a particular company, learn about particular topics much faster.

They still need to be experts in their particular topics, don’t get me wrong. But they are able to do more in their roles. And therefore there is a more level playing field being established in organisations. That really changes the traditional “pyramid” hierarchy of many organisations.

Ash: I think I’d agree with that. One of the big challenges in AI is the fact that it hasn’t been around for very long, particularly Agentic AI in its current guise. So in many of the industries – law included – you’ve always had that model where for decades or hundreds of years, you’ve had people who have become more experienced, become more senior.

How do you magic that out of thin air? Because you can’t find somebody with ten years of experience in these areas, for example. So you somehow need to be able to instil that sort of senior understanding or ‘wisdom’, for want of a better word, at a junior level and accelerate that. There’s a lot of talk around AI being used to dumb down people. You know, that people aren’t thinking for themselves anymore. But I actually think there’s an opportunity here in using AI as a teacher as well. So in the code development, copilots and tools space, you can have it generate the code for you. That is often good code that it writes.

There’s this question around: “If you don’t have the seniority to be able to know what good looks like and guide it, then is that a risk?” And it is a risk. To your point, you need the guardrails in place, but also at the same time, it’s there as a teacher. If you don’t understand something, you can say: “Why did you do it that way?” and it will explain its reasoning to you. That’s really valuable and really important to help accelerate that.

There’s always this question that people often ask me: “Is this a flash in the pan? Is this thing going to go away?” I really don’t think it is. I think we’re looking at possibly a bigger revolution than the internet or the printing press or something along that level of magnitude, because of the impact it can have. Not just in automating or doing a thing, but actually broad ranging across what people need to do. So I think there’s a huge opportunity there for people to be able to embrace it. I think it is completely fragmented at the moment in terms of technology, different vendors, different tools, and the pace of change is shocking in terms of the new things coming along all the time, and the investments in these things.

There’s lots of focus on the frontier models, and this new LLM can do this. This one’s better than this one and all the rest of it. But I think there’s so much value to be gained from actually not using those brand new models, but actually using more basic ones to do more basic things, particularly in Agentic AI. Those frontier models cost a lot of money. Those of you with your CFO hats on, you know.

Miriam: Yes, the compute cost alone is scary.

Ash: We’ve been able to engineer some of that cost out of it by using LLMs at the appropriate level. So to some extent, this stuff will become cheaper over time, for what’s required. But I think the scope of the areas that it will be involved in is just going to widen.

Miriam: 100% and you made the point about people continuing to learn and to use it in ways that are not just focusing on specific tasks. Currently, we’re working with an organisation and helping their senior leadership team to develop little mini leadership AI teams. And that is not to replace any single person; it’s a small-scale project that’s helping these busy business leaders to be better prepared when they go into their monthly meetings. Anybody who’s been in business knows that you get your papers, you try to skim them half an hour beforehand, and then you have a long discussion and then the meeting overruns. We’re changing that and helping them to make better and faster decisions, because they had the mini AI team not just briefing them, but also helping them think through the questions from very critical perspectives.

One, for example, is always the CFO perspective. That’s a small ask for AI. The compute costs are not very high at all, because it’s so small, and yet it has a measurable impact on the quality of the work that gets done by these people, in the way they can truly interact with each other and, therefore, take the business forward.

Ash: Yeah. Whilst it’s developing at such a rate of knots, whilst you know that frontier models of today and tomorrow will become the more mediocre and cheaper models in terms of that performance. I think it still does come down absolutely to working out which horse you’re going to back, because as an organisation you can’t be flip-flopping on a weekly basis, when something new comes out and have to throw it away.

So I think it’s about possibly so much change going on, in my mind, I can’t see a point in the future where those hyperscalers – the likes of AWS, Azure or GCP – will be superseded by something else. And I think that’s becoming clear with AI, is that you need the compute, you need the data, you need the storage. You can’t build that on your own, unless you have a really niche case for it, I guess, and a huge amount of investment. So whatever does come along is going to continue to be built on those.

Miriam: Yes, for sure. I think it’s a bit like: nobody tries to build a brand new search engine, right? Yeah, there are some niche cases, but we all know why ‘googling’ has become a verb by itself. So, where do you actually get the competitive advantage? That is not by building your own. I had clients who thought 6 or 7 years ago that they should be building their own. Most have abandoned that idea.

I’ll just go like, “let’s pick a platform, something that we really trust, where we can work with great partners and stick with that, at least for the foreseeable future.” That makes sense.

Ash: Maybe in a couple of years it won’t be such. Maybe it’ll just be on LLMs.

Miriam: Well, as I said, I’m staying away from the actual technology prediction, because maybe we’ll do it all by mind transmission in the first place.