AI and Managed Services

morne laubscher ai post

South Africa, Sep 4, 2026

AI is already reshaping how managed services get delivered, from event triage to first-line support. But the pace of change is exposing a gap between what's technically possible and what most environments are actually ready for.

We asked Logicalis South Africa CTO Morne Laubscher where that gap shows up most, and what it means for systems integrators trying to keep pace

.

With the pivot towards integrating AI with managed services, where is Logicalis actually running agentic AI or AI workflows in production today versus still piloting it?

 

“We've got an AI ops engine that runs on our ITSM platform that we use to automate clustering of events and triage of events.

“And that is an extremely efficient application of AI in our managed services world, because it really helps us, number one, pinpoint problems very quickly, because we don't need to sift through multiple events manually.

“The events are clustered if they are related, which makes our engineers more efficient. Secondly, it also cuts down 70% of the noise in the system, so our engineering teams resolve problems quicker, which makes the service more efficient while making it more cost-effective.”

 

What’s a realistic efficiency gain you’re seeing with AI in managed services delivery, and how does that compare to what vendors are promising clients?

 

“With agentic AI, we are now in a world where technology providers can have fewer senior-level skills and can engage agentic AI in natural language to troubleshoot and resolve issues in infrastructure and estates that we manage.

“So essentially, a junior engineer can engage an AI chatbot in natural language and request that it analyse configuration errors or gaps in network infrastructure deployments. They can give that bot the instruction in natural language to write code to remediate what has been discovered across the estate.

“We’re seeing junior resources being augmented by AI, allowing them to deliver a more advanced function, which makes the service more cost-effective and efficient.“

 

Are South African businesses building AI strategy around local data centre capacity, or defaulting to hyperscaler regions overseas? 

 

“The data sovereignty discussion differs from vertical to vertical. Not all verticals have the same approach.

“It also depends on the quality of your data.

“We are seeing customers deploying hybrid architectures, which means that they’ve got data in structured and unstructured silos across multiple technology stacks in their organisation. They are augmenting it with AI engines and processing capability that the hyperscalers provide.

“So, in other words, the data resides in your ecosystem, but the AI accessing and processing the data resides in the hyperscaler or public cloud infrastructure.

“Building AI infrastructure that processes data at scale is a very expensive exercise, so it makes sense to use a hyperscaler AI in the public cloud, but it isn’t always an easy task to move your data to a hyperscaler quickly because not all data is properly managed, catalogued, and structured.

“The other thing to consider is security and access to data. We always recommend that when you are looking at your AI strategy, you focus on not creating technology silos and having a resilient security approach before you open your data to the AI systems running in the public cloud.”

 

What's the most common readiness gap you see when a client wants to deploy AI but their environment isn't built for it? 
 

“AI POCs work well because it’s a controlled data set and all the governance and security is in place for that POC to succeed. But the common readiness gap is when that POC needs to scale across all the data in the organisation.

“It is exposed to a less clean data set with less structure and fewer governance guardrails. So what then happens is that the use case fails, even though it worked perfectly in a controlled environment.

“In our 2026 CIO report, we talk about how that is the main reason why 85% of POCs in AI fail.

“The biggest gap in readiness is doing the POC on a controlled dataset without having the right security or data management principles in place when you try to scale it. That gap normally makes AI adoption fail.”

 

Are clients asking Logicalis to consolidate their security and infrastructure stack onto fewer platforms, or still buying point solutions per problem? 

 

“The tool sprawl has happened, and organisations are facing a tremendous problem on a budget level in terms of what has been spent on sporadic and mutually exclusive tool sets.

“There is definitely a use case for consolidation in bigger vendor portfolios that cater for multiple solution sets. Examples of this would be Cisco, Palo Alto, and Microsoft.” 

 

If you had to bet on one part of managed services that looks completely different in 3 years because of AI, what would it be? 

 

“Service desks and first-line support.

“There won’t be any human beings delivering the service. It will all be AI-driven chatbots.

“It’s already happening.

“Another thing would be that the skill level required for deep technical remediation will radically reduce because agentic AI allows us to execute enterprise-wide remedial actions in natural language.”

 

The AI readiness gap

 

"AI is already delivering measurable gains in managed services, from the 70% reduction in event noise Logicalis has achieved through automated triage, to junior engineers now handling remediation work that once required senior-level skills.

"Over the next 3 years, we can expect to see that shift accelerating.

"For South African businesses, the path to digital transformation through AI requires readiness and stronger governance.

"The technology to automate service delivery and augment technical teams already exists. Success depends on whether the data, security, and governance foundations are in place before the AI is asked to do the work. "

 

 

Related Insights