00:05
so yeah uh forming a well gen AI strategy uh for a digital agency uh so yeah welcome everybody um I'm just going to quickly go through you know some of the details of what I'll be covering today I've only got 30 minutes initially I thought it was 45 so I'm going to be going pretty quick uh after doing a little bit of shopping um just going to give you a bit of an overview of how see AI is changing digital agencies um we look a bit at the Strategic imperative um and then a journey through kind of creating uh an AI road map
00:40
Discovery processes touch a little bit on some of the Practical um elements of implementing change and then wrap up with some key takeaways um so I think everyone can see like the landscape is Shifting pretty rapidly a is really changing the way that users interact with content um so I'm going to use are starting to replace Google with chat GPT um and even if you look at Google Now um you're seeing an AI summary uh of content just popping right up at the top how people accessed content online 2 years ago is going to
01:14
be quite different to how people are accessing content 2 years from now and so agencies that are harnessing these Technologies uh can deliver a lot more personalized uh efficient and Innovative experiences um so this context is really essential uh I think for for agencies like ours in understanding why an AI strategy isn't just an option but it's a real competitive necessity um so if you want to survive and actually Thrive over the next few years you really do need to prepare for where the puck is going and
01:44
not just where the puck is right now today um so let's dive into why AI is relevant so quite a few years ago at this very start of the internet boom an old boss of mine was working in ISP sales um and he met up with this guy who was like fully bought into the internet hype absolutely no idea what he was doing didn't even know what the internet was and he was like can I get an internet and he was like definitely expecting some like box or something to turn up and he was going to didn't even know what he was going to do with it had no
02:18
idea why he wanted it you know um but everyone else wanted one so clearly he had to have one right so why am I telling you this and I see a lot of parallels between this and AI um democratization right now there's a whole bunch of hype there's a lot of fear there a lot of random unstructured activity people going can I get an AI um you know and other people also going AI doesn't do anything useful um and I think both of those responses are just on two ends of a spectrum right fully Bor into the hype or fully Bor into the
02:51
fear um but then no real idea of like what are you doing with it like what problem are you actually going to solve um so jumping on the hype bandwagon it's not going to do anything for you being scared of AI isn't actually going to do anything for you either you actually need to know what are you doing with AI right how is it going to help your business what problems are you solving so whilst I'm saying you know AI is right at the peak of the hype cycle I mean if you look online at articles like
03:25
everything is kind of proclaiming things from the heavens like aen is dead aen is the next thing if you're not using it now your business is dead um deep seek you know just flew to the moon while making apro Martini um so many articles are just promoting complete nonsense so if you see headlines like that on an article most likely it's written by someone who thinks AI was invented in 2022 doesn't really know what it is it's clickbait move on there's lots of really sensible writers out there um and
03:57
there's stuff that you can read that talks about AI as a sensible tool um but yeah move on from the clickbait headlines um but saying that yeah AI isn't just hype it's not just a fad um and i' likened that to the birth of the internet because it is going to be a really massive change uh it's not just a temporary shift so we can see AI kind of getting integrated into everything right now so Alexa Siri those kind of tools hope play they actually be useful soon um um and something to remember like chat GPT 3
04:34
was released November 2022 it's just over 2 years ago so back then whenever you saw an AI photo or video had like multiple hands and six fingers and the videos would morph into weird creatures and stuff you can type a single prompt now and get a video of incredible quality that's almost indistinguishable from reality so things are progressing at an astounding rate we've got agentic systems now that can operate in fully autonomous mode doing a whole range of different tasks multimodal systems that can do voice
05:08
text images are going to be common place very soon in many businesses um we are seeing a bit of a plateau I think a lot of people following AI uh in the acceleration of model performance so lots more parameters going in um but you know the actual performance uh improvements aren't really kind of going up like they used to over the last couple of years but I would say that's most likely to drive a lot of the big business spend away from that kind of hype idea of more and more and more parameters and into
05:40
actually creating useful tools and we're seeing that now so open ai's deep research um man out of China just you know a week or so ago um and those useful tools are going to be the things that actually increase the drive of that change uh in how people actually interact with it um so yeah big question I think for everyone is like are you actually ready uh to start integrating AI into your business so um yeah so there's kind of two main things I see for most digital agencies here so it can actually enable our
06:17
customers to deliver a lot more personalized experiences using natural language um will it be perfect and will it be right all the time no it's not always going to be right uh I think there's a lot of evidence around world right now in some very big uh places that people tend to care more about fast and confident than than right um so we are going to see that kind of change in in how people are interacting um with content the really important thing for us I think is like we build websites right so that change in the way that
06:52
people create and write and manage content is going to have a really major effect uh on our businesses um a second Advantage here is really the operational efficiency inside inside your business um so for code generation business analysis testing there's all areas of tech that are really sped up uh considerably by AI if you're using it well um but not just in the tech side of things so when you're talking about policy creation um even forming your own strategy uh AI can actually help you to speed that up uh
07:24
quite a lot as well um like I said before I'm saying you know a lot of people as well still just saying I've tried a little bit of my business it's still just not working and concluding very quickly AI is completely useless it's just not a tool for me um I do really feel like this is not actually spending the time to really understand what the tool is for it's not a human it's not replacing a human today it is a tool um that you can use if I grabbed three chainsaws and started juggling with with them and lost my hand
07:58
I could conclude that chainsaws are no good if I chop down a tree with it I'd be like yeah all right I get it this is actually a useful tool um so yeah by integrating those AI services or AI into our services uh we can kind of stay ahead um of of the curb and like just really keep up with other businesses so something we are going to see is other businesses are going to be doing this um it is going to be making the way that they do business um faster more efficient they will be able to deliver more quickly so if we're not
08:33
starting to think about how we integrate this stuff into our businesses now are we actually going to be able to compete and keep up uh and that is a big question that we need to answer I really see Drupal and the open source Community having a really big opportunity to thrive through this um if you see the guys from freely give and all of the work that they're doing on on Drupal CMS and the AI initiative there um there's a lot of advantage to us actually embracing this technology and properly embedding it in what we do
09:01
uh so now onto some of the boring bit um oh sorry not winter but legislation is coming um Australia has been really slow on legislation realistically but that's probably not going to last um a lot of other jurisdictions have uh already moved so particularly the EU as you'd expect um and yes look the US is kind of relaxing some of the AI laws over there but that's not going to affect you know real world scenarios where AI um chatbots give people bad advice and lawsuits are the conclusion out of those
09:35
so getting your compliance Frameworks in place early is going to be really important um it's not as fun as writing the new app but the policy is going to be a necessary evil um hallucinations are real um transparency can actually be really tricky when you're implementing AI so you need to know how do I actually deal with those business um with those issues in my business and in the tools that I'm actually building um so from federal government now we've got the eight ethics principles already we've
10:03
got the 10 voluntary safety standards there are some discussions about nine of those potentially becoming mandatory safety standards at some time potentially even this year um and there are the mandatory safety standards already for high-risk applications um so if you're not ready for the compliance side of things um you may well build something find out that it doesn't actually comply with some of the things you need to comply with and then you're going to be sitting there doing a whole bunch of re-engineering work that you
10:28
could have avoided so oh hang that was another one of those um okay so I might be telling some people how to suck some strategic eggs here um but as I said I'm seeing a lot of people lost in hype and the fear so I kind of just focusing on some of the fundamental things of um if you were going through any new business initiative right um so assessing your current state and your um capabilities and what your Market opportunities are we're not all turning into anthropic or AI open AI tomorrow so where is your
11:03
business right now in relation to AI are you an early adopter are you a lagard and you're going to be answering these kind of questions for yourself so be honest with yourself where are you your business your staff um and really understanding what is your starting position and then going through and actually asking yourself okay well what unique value can AI bring to my agency uh because that's actually the important question you need to be able to answer um then aligning that with your business priorities so you don't want to
11:39
kind of start going off on some brand new tangent that you're an AI business so you want to have a look at what is it what are my current business goals how does AI actually match up with those business goals so is that driving more Innovation is it improving customer experience enhancing my operational efficiency and then mapping AI tools into those different processes so that you actually know what you're going to be doing so yeah for a lot of us our business um missions here around building you know well-designed and and
12:12
well functioning websites so AI goals around better CRM interfaces is probably not something that we want to do get that alignment early on about what is it that you are actually doing um you've only got so many resources in your business so you want them to still be focused on delivering for your business not just a random idea right so keeping on going with this point probably several times here because AI it's not a silver bullet it's not actually going to fix anything for you you need to get in
12:41
there and understand what you're doing with it and how you're going to be using it um next setting some measurable objectives so kind of oh sorry breaking down that Vision into some actionable objectives so if that's you know um objectives and key results so okrs or kpis if you use a strategy house in your business um go through and actually get some definition around um around those objectives and how you're going to be tracking them um once again just aligning that with the way that you measure and plan out strategy in your
13:15
business right now today rather than starting to think oh hang on once again this is AI I'm going to go off on some random tangent um align your new AI initiatives with how you measure success right now today um so yeah plan out your goals and then plan out your success measures like how are you actually going to measure the success of these new objectives that you've got because that's going to be really key um to what you're doing um this another really important one but actually fostering an innovation culture
13:47
inside your business um it's going to be the people who are actually building these new tools and integrating this new stuff into your business the AI isn't going to do it all by itself um like everyone here today um you're going to have staff in your business who completely believe in the hype or they're completely lost In Their Fear or somewhere in the middle um so it's going to be on kind of the leaders of your business to set what that strategy is and then start to Foster and drive that
14:18
culture inside your business okay um so yeah so now that we've kind of got some of our high level uh strategy items all kind of laid out we kind of go through and actually start making that more practical so in this phase you want to kind of get a clearer picture of um of where you currently stand so it's taking that high level thinking from your objectives and starting to add some more detail um so begin by really kind of going deep into assessing your AI capabilities what are your data assets what are your current
14:52
operational uh processes and how you're changing them um so this assessment is going to help you identify those key gaps um and uncover those really high impact opportunities where you can really make a difference with AI um and so this is kind of really laying that Foundation that you're going to spring board off from so once you've got um a solid graph grasp of where you are um and some of those opportunities uh you want to kind of start phasing those out into some actions that you're going to actually
15:23
take so you know dividing those into three segments here of just some short-term Wis um so some smaller projects that are relatively small investment low risk get you to actually start practicing um embedding these things inside your business so an example of this could be starting to get your Dev team uh properly using Code um aib based code editors um on an everyday basis um some midterm projects right so these are going to be some of your um bigger initiatives uh that can go out on a bit of a broader scale across your business
15:58
um so this is kind of really refining the approach to how you're embedding AI um scaling some of the successful Pilots um and integrating those into your business processes so an example here might be building an internal chatbot uh that has access to your um internal knowledge base to help your staff um find information faster and easier and then some of your long-term Vision items so obviously these are the bigger transformational projects that you're going to um be embarking on um are you becoming an business actually
16:30
providing AI products or is this all going to be internally focused and more around efficiency um so these are the really big ambitious initiatives that kind of align with those longer term strategic goals if you have got a whole bunch of ideas um using some form of Matrix doesn't need to be this one there's a whole bunch of others online um online that you can look at um in this one here it's basically how much value does this idea have for your company or your clients and then how much effort is it
17:02
going to take uh to actually go in and create it um obviously build your quick wins items um up in the top there as as you pilot and as you start to feel more confident you kind of move into those bigger strategic uh projects something really important I think in this phase is don't be scared to put ideas in the bin if something is going to take a load of effort and it's not really making much difference to your business no matter how passionate you are about it um even if it's your own idea you need to be willing to just go actually
17:35
you know what I don't have enough resources in my company it's going in the bin um okay so essentially now you've planned out the work uh now you've just got to work the plan okay so um implementing AI isn't just a technical change as I said it really is a people challenge right so you really need to ensure that the teams around you uh are ready to implement this this change um so find the people in your business who really are your uh kind of change Champions um and provide them with those
18:09
opportunities to grow um so this includes kind of targeted training and upskilling for some of those people but really creating an environment where that implementation and that Innovation is really encouraged um and effective change manage management obviously requires clear and consistent communication um so your staff need to see that those who are embracing the change are rewarded um and have those kind of incentives to further develop what they're [Music] doing okay um so up above we've kind of defined the pilots in our road map so
18:48
essentially go out and build them um in Your Vision objectives you would have defined your measures so use those measures now and actually keep track of the work that you're doing uh governance yes once again the boring bit uh but while you Tech folks are off doing the fun building stuff know there's actually other people who really do love policy sorry um but yeah make sure that you're actually uh out there and tracking and embedding the governance rules right so the mistakes that you can make here are by providing
19:21
bad governance to real customers can have real world financial problems for you in future um so don't ignore the boring stuff and make sure you actually operationalize the governance so it's not just about creating new policies of shelfware it's making sure that everyone in your business understands them and that you're actually embedding those things into the things that you are building um and then start to scale um so use the measures um that you had before the things that are working obviously scale those things up um try
19:58
to like retain ownership of those scaling up projects to the cross functional teams you use to actually build them so that you get that kind of ongoing interest and and ownership over the things that you're building um and don't forget to go through and actually do your Lessons Learned um processes as well so that you learn from the things that you've already done okay um and then a really important one I think a lot of people forget um just like any other thing that you do don't forget about resourcing your Bo
20:29
activities right so the new AI toys aren't just for Christmas the new training goals are not just once off the new policies aren't set and forget you actually need the time to go back and maintain all of these things the AI space is going to be evolving quickly over the coming years so actually planning to continuously resource and change in this space is going to be quite important so s of digitals Journey on this so far um early on we also kind of affected by a lot of excitement and um you know maybe not enough focus and
21:05
we're really tackling this now like any other new line of business right so if we wanted to move into building Magento sites it wouldn't just be a random activity if we were embedding iil processes across the business once again we'd be taking this kind of structured approach um a big thing for us because we've got such a strong government Focus um we really do need to try and avoid public llms as much as possible and focus on kind of self-hosting and other things like that that's not going to be
21:33
the same for everyone's business um and that's what I've been saying all the way through the important thing about this strategy is going through the process and doing the thinking for you in your business how are you actually using these tools um training and sharing knowledge has been really important we've already identified some kind of change Champions within the business um and providing some spaces for those people to actually you know discuss those changes with um uh with people right across the business
22:02
um right now we're kind of focusing on building some internal tools um really starting quite small uh at the moment um and yeah focusing that attention mainly on the tech team and how we actually kind of in uh integrate AI within the technology that we're that we're building um so I'm going to flick through these slides really quick um as I said the uh time was a lot shorter than I actually actually thought you will have access to the slides later and most of these are actually links that you need to read offline uh
22:37
anyway um so we've got some Readiness checklists and maturity assessment links in here that are really good to just go through to give you an understanding of some of the things you should probably be thinking about uh when you're looking at an AI strategy to really understand your business um when you're selecting your tools doing a proper assessment is really good this is actually quite a good vendor assessment tool here if if folks want to use it um there's some guidance here on kind of building an AI
23:03
Road mapap um which is quite a handy tool as well um and I've got some links here on kind of um guidance around staff development and things you need to train your staff some people will interested as I said some people are going to be scared everyone's going to need some form of training um the people who are kind of hype uh zealots are probably going to need some compliance and transparency training and things like that the leg guards are probably need some training on how to properly use the tools um once again compliance is going
23:34
to be key I've got links in this page to all of the different um compliance things that you should probably be aware of it's really just to the front page of those things this is government legislation here so can be a very deep dive if you go right into it um okay and then just yeah some more links here again so like some of the key takeaways from today um I know I've covered a lot very very quickly um so you know the key message I want people to take away is AI is a new tool right it's how you use it that actually
24:09
matters um and this is on you AI isn't going to do this for you it's not going to implement itself you need to take action and actually go through and figure out what am I going to do with it um so as I've said yeah many businesses really getting caught up in that um hype um and not going through that real planning phase of understanding what action they're going to take um success as in with any other business is really going to be going through that planning aligning those AI objectives back to
24:42
your business objectives so you can actually drive some success um so if you are thinking about AI in your agency starting with that Readiness assessment is a really good place to start um and then once yeah we've to find your opportunities go through and actually just get started on your um on your Pilots so I'm getting a 5 minute wrap up here so go onto ah those were meant to click through those as I was talking well done me anyway they're up there now um so I could tell you more but here are some
25:24
quotes from some very wise men who know a lot about AI um over to you folks for some questions I think you wonder how far are you down the pathway measuring efficiency and processes go to be honest that's been really difficult so we're focusing more on kind of qualitative qualitative assessment there so like surveying staff um before we kind of introduce new how are you actually implementing or how are you doing your business right now how long does it take um checking if people are already using AI um and then doing
26:09
assessments afterwards and then just looking at the difference we haven't got any really good kind of automated or quantitative measures at the moment yeah measur exactly it kind of depends um on how mature your measurement is of efficiency right now so the more mature that is the easier it is going to be for you so yeah yeah so yeah you have a problem with like people saying oh I need this tool and you're saying well we've got a strategy around this platform so have you actually really checked it um I was in one organization
26:45
where I a conversation large organization not here and they said there was going um we can't just keep buying subscriptions for everyone just because they Lear that they can do on the next AI tool we need to actually mature around platform so are you balancing that are you trying to mature around my platform where you kind of actually we don't know so at the moment so for um Dev tooling we are kind of focusing on cursor a lot at the moment but not then preventing people from using other things but uh sorry y but yeah uh cursor is
27:30
basically a reroll of vs code um so it's a coded edor editor with um a chat agent built in yeah so actually using a range depending on um what activity people are doing so what we've been focusing on is actually getting really good Confluence docs around here are some good ways to set up your initial rules um this is how you operate with it when you're asking these questions maybe use um you know Claude three dots seven thinking if you're asking for architecture if you're going to ask for code use C on at 35 and
28:05
providing more direct guidance um because I think you need to give people the training on exactly how to use the tool which you can't really do if you've got a really broad spread but then as I said we have other people if you got licenses for every PL person yeah exactly and then I think you're not going to see the efficiencies out of it as well think that would beow Tok yeah yeah and with Cur you can actually hook that up to your own apis and stuff as well if you want to say so yeah yeah um other questions guy
29:03
going okay so that question was around had you actually manage transparency or compliance and transparency essentially um I would say it is actually really difficult um but you can actually control when you're using different models how creative that model gets um and I think this is where you're seeing a lot of people move to more multi-agent systems so instead of trying to say I'm relying solely on chat PT to give me the end to end answer for a problem you're going to see a chain of Agents where
29:34
your initial question might actually go to chat jpt and then you'll have a secondary agent that's there that is not going to be very creative and it's going to go back to Source data and check the answer that came out of agent one and say is this actually accurate and honest um depending on the results you get you might actually have two three agents going through doing some editing and and verification um but you can actually do that in your instruction sets as well right so in the base instruction you can
30:04
say do not give me anything that you have just created go back and make sure that you have actually verified that links exist um that the things you are saying are based on documents that we have and and those kind of activities So It's Tricky but it's completely possible and I think you're seeing that coming out of some of the multimodal or multi-agent systems that are getting built now as well other questions one more around training comp to teams um we have a safe responsible new AI policy we do training you do on
30:40
boarding um but continually see people kind of tripping up um and making mistakes um wondering how how you would handle ongoing of knowledge so I think this is is why I kind of kept reiterating that AI is just another tool because I think you'll find the same thing with your cyber education um you're going to find the same thing with your HR um rules and policies around your business it's just another tool in your business there are going to be people who read the policy understand it comply with it on day one there are going to be
31:17
people who fail um so you do the same thing you would do with every other policy you probably redo your annual training um if people are repeat offenders you go back and you have a conversation with them and remind them how to do things um I don't think there's any kind of Silver Bullet around compliance um it's the same old way people have been doing it for a long time yeah all right thanks very much kit thank you
Reviewed 03 June 2025