In the first article in this series, about a $1.6m lesson, we looked at how AI can start to resemble shadow IT, with lots of sensible local decisions, but little shared enterprise capability.
It’s not just Shadow AI that represents the risk here. There’s another force accelerating that problem. Right now, everyone wants to be involved in AI.
And that’s understandable. Nobody wants to be left behind in the AI revolution. (Almost) Everyone wants to upskill.
No executive wants to explain in two years’ time why their business unit sat on the sidelines while the AI revolution happened around them. Teams want experience. Leaders want visible progress. People want to understand what AI could mean for their jobs, customers and operations.
That enthusiasm is as palpable as it is valuable.
But it also creates a risk closely linked to the Shadow IT pattern we observed previously.
.
When every business unit wants an AI initiative, it’s easy to measure progress by counting projects.
Ten pilots sound better than two.
Twenty use cases sound better than five.
But an organisation can have dozens of successful AI experiments and still be poorly equipped to deliver AI at scale.
There’s almost no question that the project is likely to have been helpful for the people involved, giving them valuable experience working with AI. The question also isn’t just whether the project worked or added some sort of additional business value.
Instead, it’s a question of whether each project left behind valuable foundations for others to build on.
Did it establish reusable data models, data access or improved data quality? Did it create integration patterns others can use? Did it improve security controls, model evaluation, observability or governance? Did it capture knowledge that prevents the next team repeating the same discovery work (like those chatbot discoveries in yesterday’s story)?
If foundations aren’t built for others to leverage, the next AI project may start almost exactly where the previous one started.
.
If we take this perspective, it changes how we think about AI return on investment (ROI).
In these nascent stages of AI implementation in telcos, the value of AI project #1 isn’t only the benefit delivered by that individual use case. There needs to be centralised implementation planning such that part of its value should be how much cheaper, faster and safer it makes projects #2 through #20. Contributing to something bigger than itself.
That doesn’t mean every AI experiment needs to become an enterprise platform project. Not at all.
There are literally hundreds of possible use-cases where AI can enhance OSS/BSS workflows, tools, integrations and more. The challenge is trying to identify and prioritise the use-cases that have greatest impact at moving the needle for the business.
Part of the prioritisation assessment is in deliberately identifying which use-cases or parts of the work are reusable, then make them available to others.
Some teams will contribute data patterns. Others might establish models and model access, APIs, plug-ins, policies / controls, operational processes, governance models or lessons about what does or doesn’t work.
Over time, those pieces should accumulate into capability.
However, from what I’ve seen so far, it seems that there are many decentralised AI projects underway within client organisations without having a centralised, coherent view of how those resource investments are cumulative.
.
The wrong response would be to put a central approval gate in front of every AI idea.
That’s silly because each business unit understands their context, stakeholders / customers, constraints and opportunities better than anyone else. We want to encourage them to continue experimenting.
The challenge is figuring out a way to create enough coherence that their investments don’t remain isolated.
The goal shouldn’t be fewer AI projects. It should be an organisation that gets smarter every time it runs one.
That becomes even more important when AI moves beyond a contained pilot and starts interacting with the production systems that actually run a telco.
In the next article, we’ll look at why almost every serious telco AI initiative eventually collides with OSS and BSS – and why AI, Autonomous Networks and Autonomous Operations can’t be planned entirely in isolation.
Download our AI / AN / AO flyer for an overview of the challenges we’re exploring throughout this series. At the end, we’ll bring the different perspectives together into a practical take-home pack for transformation planning.
AI / AN / AO (Artificial Intelligence, Autonomous Networks and Autonomous Operations) transformations create different challenges for different stakeholders, depending on where you sit in
Our last article asked a simple question, “How much are you effectively paying to stay on the antiquated side of the Buyer-Seller Chasm?” Let’s be
I’ve been trying to teach my kids a lesson lately that, if I’m honest, I’m probably not succeeding at (yet). Recently, it struck me that
The previous article highlighted how my chiropractor taught me something I didn’t realise about telecoms. It then triggered another image in my head (the first
