Subnet's 'cost neutral' managed services long game

Subnet managing director Brett Lodge is looking ahead to what the MSP should look like by 2030.

Today, the 25-year-old Adelaide-headquartered firms’ work would look familiar to many MSPs, but Lodge is gradually reshaping how it operates and what it sells.

This includes building an AI and automation team in Sri Lanka and aiming to make managed services “almost cost neutral” for customers by the end of the decade.

In 2017, Lodge set a long-term target that by 2030, much of the business’s activities should be automated or self-service.

“So basically, 90% of what we were doing in 2017 had to be gone, and we had to be doing something completely different,” he said.

When we spoke with him late last month, Lodge estimated that 35-40% of the business had been automated.

One of the areas Subnet had targeted was onboarding new users – a task MSPs can still chew a lot of time on.

“We do a lot of news or onboarding. So we saw a brand new user taking 44 minutes while we're doing it manually, dropped to eight minutes. Now we've got tools where we can integrate that with smart forms to take it down to zero minutes to do news or onboarding,” he said.

“Now, we don't necessarily pitch that as a massive cost saving, because it is that, but it's also about getting things done perfectly every time, and not having the security issue of someone accidentally putting the wrong group or doing the wrong thing.”

Subnet has leaned on low-code tools and people inside the business who want to specialise.

“Because we don't have coders, we're using things like UI path, which is much more of a WYSIWYG dragging drop kind of automation tool,” Lodge said.

The next step is to bring in dedicated automation talent, with a potential source of skills in Sri Lanka, where major vendors have invested in people.

“Microsoft and everyone else has invested a ton over there. So being able to leverage resources are already on the ground. It's going to be where we pick up over the next year or two,” he said.

Lodge expects AI and automation to strip out many low-level service desk tasks over time, particularly as phone-based tools and self-healing features improve – but he said the intent is to shift those roles rather than remove them.

He wants front-line staff to become “supremely customer focused”, handling the human interaction work while automation and agents take care of repetitive fixes in the background.

Alongside that, Subnet has been moving from best-of-breed point products to what Lodge calls “Subnet-based services powered by whatever the vendor happens to be” –  aiming to put its own processes and reporting at the front, and swap tools underneath without customers noticing – or eroding margins.

Managed foundations, security + “insight”

On the customer side, Subnet is structuring its offer into three named layers: operational support, a security layer, and a future third AI and automation tier.

The third tier will be aimed at customers wanting to run their own AI models and automation internally – and have their MSP support that.

It assumes some customers will not just consume cloud services, but will build their own “AI factory” on top.

“They're going to have an AI factory, effectively, with their large language model internal. So how do they manage that? And how do they support that? What does that look like? Because that's very different to a we put a remote management tool on a device. It's actually having to manage the information.”

“So that's going to be the basis of where it starts. And this is kind of in the 2027 timeframe when that's kicking off. We've started to work through the bones of that.”

In practice, Lodge said many smaller customers are still grappling with basic security governance, while mid-market and larger organisations “get it and really need help and support with it”. So Subnet often starts by auditing existing protections and putting “guardrails of governance” around security and AI use, giving clients clarity when it comes to tool selection.

He is seeing early versions of the “AI factory” idea with clients that have large, sensitive data sets and are looking at their own AI infrastructure because “they can’t have that data go external”. But he expects those on-prem large language model builds to remain relatively niche.

From best of breed to “Subnet-first”

Subnet is also looking at the profitability of its software stack.

Previous successes include the adoption of Kaseya Quote Manager. This gave customers greater visibility of available stock, which Lodge said resulted in an increase in margins.

“We saw customers that were very price conscious and always shopped us around 1,200 times to different vendors go, ‘Okay, we'll just actually order everything from the store’.”

“And so what we're trying to do now is do the same thing across different stacks”.

On the security front, Subnet previously moved away from emphasising best-of-breed security brands to customers – it now pushes “Subnet-based” services, with vendors swapped in behind the scenes.

“We're getting to the stage now where it's less about the product and it's more about us as a vendor. That maturity is allowing us to pick tools that we can actually make more margin with.”

Subnet isn’t using Kaseya “across the board”, but Lodge saw merit in what he saw as the vendor’s “all-in-one” approach which he saw as lowering the cost of tooling – then “your focus is your profitability based on being able to deliver those services”.

Making the numbers work

“By the time we get to 2030 is, I want to make a managed service that's almost cost neutral to a client,” Lodge said.

The thinking is that a managed service might cost a customer, say, $15,000 a month, while the customer’s data and AI agents might generate closer to $50,000 of operational value.

“They're getting 50 grand of value for a 20 grand-a-month managed service – that's a position we really want to get to,” Lodge said.

“Then it’s far easier to have that conversation with a client – spend small amount of money, get a large amount of return. That's very different to the model today, which is, ‘I'll support your thing, and there's 100 and something dollars a month.”

“That's our slightly longer-term plan. Obviously, there's a lot of work to do in between now and then, but that's what we're scaling up for, to be able to drive that.”

techpartner.news was a media of Kaseya Connect APAC 2025.