AI training tops planned investments as agentic AI scales up

What happened

A new survey from Liferay finds that AI training is now the single most commonly planned investment among organisations over the coming year, as agentic AI moves from pilot projects into more structured, organisation-wide rollouts.

The finding points to a shift in how businesses are approaching artificial intelligence: rather than treating it purely as a technology procurement decision, organisations are prioritising the people side of adoption — building internal capability and literacy — as agentic systems take on more autonomous, decision-making roles within workflows.

Why it matters

The prioritisation of training over other categories of AI spend signals that many organisations recognise a capability gap between the tools they are deploying and the readiness of their workforce to use them effectively. As agentic AI systems increasingly act on behalf of employees — making decisions, triggering workflows and interacting with customers — the risk of poor outcomes shifts from a purely technical problem to an organisational and behavioural one.

For leaders overseeing digital transformation, this suggests that success with agentic AI will hinge less on model selection and more on how well employees understand, trust and correctly direct these systems. Training investment is, in effect, a proxy for how seriously an organisation takes the operational and governance implications of moving AI from experimentation to production.

The Renascence take

It is telling that training — not infrastructure, not new tooling — tops the investment list. This is an admission that most organisations' AI ambitions have outpaced their teams' ability to safely and confidently operate the technology.

Treating AI training as a line item misses the point: the real work is redesigning roles, decision rights and escalation paths around what agentic systems can now do autonomously. A customer-obsessed operator should ask not "how do we train staff on the tool" but "which decisions should a human still own, and why" — because that boundary, more than any course, determines whether agentic AI improves the experience or quietly erodes trust in it.