CRN Australia recently spoke with Marc Tricksey, sales director of Arinco, about the Fast50 firm’s work delivering AI and data projects for Australian and New Zealand customers.
Arinco had over 40 AI and data engagements in the 2024 financial year, representing a 117 per cent increase from the previous financial year.
Tricksey said he expects the firm’s AI and data engagements to “at least double in FY25.”
We wanted to get a flavour of Arinco’s recent AI and data work and the challenges it’s helping customers with.
CRN Australia: Can you give us a sense of the kinds of data and AI projects Arinco has delivered?
Marc Tricksey: I think it's twofold. There's the data side, and we've seen a big uptake in a couple of things within data. One has been Fabric; so Microsoft has obviously done a lot of work to [provide] a SaaS data offering that’s a lot more simple to consume and use. Small organisations are, typically from what we’ve seen, moving towards a Fabric option to standardise their data and bring together those repositories.
The other one is to do a standardisation, but on a more enterprise sort of scale data platform. In some cases it can still be Fabric, but we've seen a lot of Databricks being used in that space as well. I think it's really…that people are starting to acknowledge that AI is here to stay and it needs a better dataset to work on; so they're starting to get ahead of the curve. Even if they're not necessarily doing much with AI, they're starting to drive more data initiatives to get the data in the right shape, the right form, the right quality and [implementing the] controls to be able to use it for AI use cases in the future.
For example, one of my customers in the financial services space, they're looking to pivot and become more of a data-led product organisation. So they provide products today, but the data doesn't necessarily influence a lot of it. We've helped them put in a data platform and we're now starting to work with them on how [they can] move data sources across into the new data platform on a product by product basis to further inform the direction of those products going forward.
So that's one trend, and the other is the pure AI side of it; it's still very much knowledge based. People are dipping their toes in with knowledge based use cases to start with because it's a big percentage of their population that want it; everyone wants to get quicker access to that data and they’re starting to understand that they can really automate some of the manual processes that are happening today.
That could be things like where you've got a compliance framework; you maybe have breaches being reported and there's dozens if not hundreds coming through per month to reporting teams. So being able to apply a simple compliance framework across forms, or whether it's emails or documents that are coming in regarding breaches, and instantly assessing their applicability or not, is the sort of thing customers are looking at going ‘we can actually save a lot of manual time here and have our people work on more interesting and more exciting things.’
CRN Australia: Were there any technical or business challenges involved with moving the customer’s data sources to the new data platform?
Marc Tricksey: The bit that we’re starting to help with now…it's more of a mindset change; so how could they use the data to inform their product’s direction, product change etc. That's a big part of it and that's where I think overall where we've been really successful in the AI space [with] our sister company D6 Consulting. They bring that business consulting experience to the table and they're the ones who are running this engagement and working with the product teams and the current owners of these systems and processes to go, ‘this is what your typical use cases look like. Here's where data can start to inform that, therefore these are the types of datasets that we want to load into that new platform.’
Because they're very conscious of not just lifting the data repositories they have today and putting them into the new platform; they want to make sure they are constructing it the right way so you can add more value going forward. So I think it’s that business understanding; it’s bridging the gap between business users using the system to business users using data to help inform those systems going forward.
CRN Australia: I guess the other end of the story is you can have these amazing systems, but if your workforce is not guaranteed to work in that way, it can all fall flat on its face. So you have to start with the business understanding.
Marc Tricksey: You do and it's a very good point. I think we've done well with our D6 Consulting partnership because they have that focus and we see better adoption when we have the engagements that are hand in hand. We're running the technical side, they're running the business side, and we're helping to find more use cases, but we're also developing a change and adoption roadmap. That covers off training, responsible usage, policies, all of the surrounding business processes that need to go with an AI implementation.
One of the big law firms we did [work with], they made a conscious decision to invest in their people, invest in the training. They knew that they needed to make space for their people to be able to spend time and immerse themselves in AI for them to naturally want to pick it up in their day to day jobs. So they came up with a pretty clever way about how they could give them more development time and not impact the billing rates etc. to make sure they were adopting, and they've seen some really good results from that.
CRN Australia: I imagine for some organisations there might be pressure from boards or senior management to see how their AI project is going. Are you seeing much of that? How are you balancing that pressure with getting things right?
Marc Tricksey: Yeah, we undoubtedly are. I'm having calls from roles like head of innovation…one I'm talking to at the moment, the board are on him regularly to say, ‘what are we doing? Why aren’t we moving this forward? I think it's having a pragmatic approach, which is generally how we go about things. You're right; it's about trying to get the balance. We don't have to have everybody upskilled to get something in and implemented from a proof of concept phase.
We're very much proponents of start small and build from there; so some core capability that you can roll out to a small pilot group and then grow from there. But it's got to come hand in hand with a certain amount of that change control, and that's where I think we’ve just got to have the honest conversation [and say] ‘it's not going to work if you just turn this on.’ I think some organisations have seen this already with Copilot. Copilot has got really good potential but it takes a bit of getting used to.
It's a human change; we’re creatures of habit and having to check with everyone if they’re okay if I transcribe [a meeting with Copilot] so we can automatically generate notes which will help us all; it doesn't come to us straightaway. So [it’s about] spending time and investing in that change side. I think people are starting to understand it; the cycle seems to have gone from ‘let's just do a proof of concept really quickly’ to people going ‘AI is here to stay and can give us benefits, let’s get it out but let’s get it out the right way.’
I think people are on that journey and it's accepted that it's going to require new changes and end user changes the same as your standard Windows upgrades or your device fleet refresh where you've got to do a good amount of change control otherwise people don't adopt the new ways of working.
CRN Australia: Arinco delivered over 40 data and AI projects in FY24. Can you give us a feel for how many are still in the POC phase and how many are in production?
Marc Tricksey: A lot of it has moved to production now. What I would say is there's less that's moved into customer facing production; a lot is in internal facing production. I think there's still a bit of nervousness about releasing customer facing AI and the typical path we're seeing is ‘let's get it into a POC, let's get it into general production internally, let's let our teams use it, let them become comfortable with it and keep enhancing it, keep tweaking it based on the learnings and then push it out to external facing production.’
ROI is a big thing on this. As much as organisations want to embrace AI, ROI is important. It comes down to the CFO standing up to fund these activities; manual time savings internally by automating these processes are significant. Again, I think there's a lot of that focus internally, especially professional services firms, legal firms, they're the ones who are still really leading the charge, and the ability to speed up and take time out of those manual processes.
There's not as much nervousness about going into production now; the technology has been there and proven over the last 12 months for many organisations. Most of what we're seeing now is that it's still going through a pilot phase but moving quickly into production, whereas in the early days, 12 to 18 months ago, it was ‘proof of concept; let it sit there and maybe not do anything with it.’
CRN Australia: Has this led to customers having more confidence to expand AI into other parts of their business?
Marc Tricksey: Absolutely…it’s that word of mouth. People are starting to use it, they're getting good outcomes and then they will go ‘maybe I’ll go to the product owner and see if I can try this with it.’
Financial services [AI projects] are starting to gather momentum and a lot of it is customer facing use cases; how do they help their contact center staff provide better, quicker information, how do they reduce the time to transcribe call notes and look at next best actions and those kinds of things.
I think the customer service use cases, there's a lot to be done. Nobody wants to spend time on the phone waiting to talk to somebody. Again, the more they can speed it up it leads to better customer retention, which is obviously a huge metric for most.
CRN Australia: Can you provide an idea of the kinds of skills needed to keep iterating AI? Have your customers needed to hire new staff or do they completely rely on Arinco?
Marc Tricksey: Part of our approach in general is we want to leave our customers able to support and develop these products and services themselves. As much as we love working with customers…I don’t want them to have to call me every time they want to do something. So we spend a lot of time working with their teams to upskill them and be ready for that ongoing journey.
What are the new skills [needed]? To be honest, it hasn't been too much of a stretch for most [of our] developers. Most of our team have got application development backgrounds; a lot of .Net, a lot of back-end capabilities. They've got the right sort of mindset to be able to go ‘what would work as a good prompt, as some good tweaks for how we interact with AI to get the right outcome?’ That’s typically what we've seen is developers stepping into this space, because it's maybe half a step to the right of what they've grown up doing anyway, so it hasn't been too much of a change.
CRN Australia: What kinds of AI challenges are you anticipating for customers in the next 12 months?
Marc Tricksey: I see customers still struggling to understand how and where the use cases will come from; I think that's still a big piece of it. Skills a little bit. Skills are always a challenge in the technology industry; the industry keeps growing at pace from what we see, so having people who can either step sideways into these sorts of roles or you're going to get the younger generations who are coming up with these skill sets first and foremost, but really I think it's around helping those organisations identify the use cases and be brave and take steps.
You can get going for fairly minimal costs these days in terms of AI, so getting in there and going with a solid plan for a few use cases to save them money…and giving them better outcomes as well; I think that's a little bit of a hurdle that needs to be overcome this year. Beyond that, I'm sure Microsoft has got different things in the pipeline in terms of not only upgrades to capability but different offers for reserving compute; so different models around ‘how you can get better performance out of this today.’
Because some organisations are definitely going to want [capabilities from Microsoft] around reserve compute for AI so that they know if adoption does start to go the way they're expecting, they're not going to run into limits and caps, they're going to be able to really, really push the use of AI because they've got this compute there on demand; potentially the ability to have that burst of capacity which is what cloud is all about. If they've got a use case that maybe only needs to run once a week to process a lot of documentation rather than paying for levels that are here, they can just burst up to it [with a] pay as you go type service.
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
