.svg){kind=small}
As 2025 comes to a close, AI remains one of the most discussed (and possibly, misunderstood) topics in CRM. Over the past year, operators experimented heavily with automation, machine learning models, and predictive tooling, all in pursuit of personalisation at scale.
What became clear, however, is that AI can’t succeed by removing humans from the equation. It succeeds by changing where humans added value.
Why traditional marketing automation fell short
Traditional CRM automation relies on static logic and predefined journeys. While effective at smaller scales, these systems struggled as complexity grew. Every new segment, regulation, or behavioural nuance introduced more rules, more exceptions, and more manual oversight.
By mid-2025, many teams found themselves maintaining automation rather than benefiting from it. The promise of “set and forget” marketing remained elusive.
Agentic AI, with guardrails
Agentic AI introduced a new model: systems capable of observing player behaviour, evaluating context, and recommending or executing actions dynamically. Crucially, the most successful implementations in 2025 weren’t fully autonomous.
They were human-guided.
CRM teams defined objectives, constraints, and compliance rules.
AI could handle real-time decisioning within those boundaries, allowing it to continuously optimise based on live feedback.
This balance matters. In regulated environments, trust and control are just as important as speed and scale.
How human-in-the-loop reframed adoption
With all the progress made in agentic AI during 2025, one reality remained consistent: trust.
While the technology itself advanced rapidly, adoption was more measured. For many operators, handing full autonomy to AI systems felt like too large a leap, particularly in environments where accountability, explainability, and brand responsibility matter.
This is where human-in-the-loop models played a critical role. Not as a compromise, but as a necessary bridge between innovation and confidence.
Rather than positioning AI as a replacement for decision-making, the most successful approaches treated it as a powerful advisor capable of analysing behaviour at scale and surfacing optimal actions, while leaving final intent, validation, and governance with humans.
A more strategic role for CRM teams
Rather than removing human input, agentic AI shifted where that input mattered most.
CRM teams spent less time predicting every possible player path and more time:
- Defining strategic objectives
- Setting guardrails and success metrics
- Interpreting outcomes and learning from edge cases
It was about applying it more intentionally.
The lesson from 2025
The lesson here is that no matter how amazing technology gets, the "why" is the most important thing. Humans are still the best bet and will continue to be. The technology is ready but the industry is still getting to grips with the realities of adopting it. It's a big disruption so it will take time for us to really see the change.
No matter how sophisticated AI gets, the “why” behind each decision still matters. And humans remain best placed to define that purpose.
