With all the excitement and hype around AI, inevitably people are asking whether it’s still worth replacing legacy core systems or whether there’s an easier AI-powered way forward. It does seem plausible that AI-based strategies could avoid core system replacement. The real question is when, and whether it makes sense to wait.
Three approaches
- AI agents rebuild your core
A group of AI coding agents could examine your existing core systems and redevelop them for the cloud. They’d need to incorporate modern requirements (products, distribution, regulation) while preserving useful functionality and removing legacy constraints.This is essentially what modern core platform vendors have been doing for the last decade(s), except your AIs wouldn’t have input from other carriers or the time for years of real-world trials.
- AI replaces your admin teams
Another approach would be to replace your admin teams with AI. AI admin teams wouldn’t need a core system; they could keep records as they needed to. This would be a bit like pre-computer human teams using ledgers and paper records, except of course the AIs would be keeping digital records of their own design to meet customer needs and achieve your business goals.
- AI wraps and enhances your existing core
You could also use AI to develop solutions to wrap and enhance your existing core solutions, so ideally nobody would notice your legacy systems.Our industry has a long history of the human variant of this approach. While you can achieve some short-term goals in this way, it leads to more complexity as this year’s workarounds become next year’s constraints.
What level of AI is needed?
These approaches would be difficult and time-consuming tasks for teams of human experts. You might even say beyond the capability of human teams in a reasonable timeframe. It’s clear the AI in question needs to have at least human-level intelligence and the ability to acquire human-expert level skills in both technology and employee benefits.
No AI available today is capable of implementing these strategies.
However, many leaders in AI are forecasting human level intelligence (artificial general intelligence or AGI) and even Artificial Superintelligence (ASI) in anything from this decade to coming decades. On this basis it’s reasonable that, at some point, AI will remove the need for core system replacement.
But when?
There are lots of experts and lots of opinions, many of the most optimistic come from the AI labs themselves. One thing we can be sure of is that this level of AI isn’t going to be available soon enough to build into a business plan. It’s more hope than strategy.
There are other major problems
Even assuming we have AGI, it’s going to be cheaper, faster and lower risk to use AI to implement a proven core platform than it would be to create, trial and stabilize a new one.
You’ll want to select one with embedded AI, in the cloud, with digital service and modern business practices. But simply the time, cost and effort of implementing an existing package, human or artificial, is much less than creating a new core system or replacing entire business units. It also has the advantage of being possible today.
There’s also a bigger problem. Any AI that can implement these strategies will be doing most of jobs being done by people today. Without people in most jobs, who needs an employee benefits industry and its core systems? We’ll either be sipping robot-delivered pina coladas on the beach or slaving under the yoke of Skynet. Either way we won’t care much about core system replacement.
Be careful what you wish for! If you wait for AI to replace your core system, you may find it has also replaced you, your industry and your interest in core systems.
