AI Complexity Emerges as CX Risk for UK IT Teams

What happened

New research from Freshworks indicates that artificial intelligence is adding to the workload of U.K. IT teams rather than reducing it, with AI-related tasks consuming budget and staff time that would otherwise go toward strategic projects. The findings echo a broader pattern identified in early 2026 research into "botsitting" — the growing need for humans to monitor, correct and manage AI systems — alongside gaps in AI governance and questions about how ready enterprises really are to run AI at scale.

Collectively, the research suggests that many organisations have deployed AI tools faster than they have built the operational structures to support them, creating a new layer of internal administration around systems that were meant to remove it.

Why it matters

The core promise of enterprise AI has been efficiency: fewer manual tasks, faster resolution, more headroom for teams to focus on higher-value work. This research complicates that narrative. If IT and support functions are spending meaningful time babysitting, tuning and troubleshooting AI systems, the technology is shifting effort rather than removing it — and that shift has a direct route into customer experience, since IT capacity constraints and budget drain tend to show up downstream as slower fixes, delayed rollouts and less resilient service infrastructure.

For transformation leaders, the signal is less about whether to adopt AI and more about how deployment is sequenced. Tools introduced without clear governance, ownership and monitoring plans risk becoming a standing operational cost rather than a one-off investment, which changes the ROI calculation many boards were sold on.

The Renascence take

The uncomfortable truth in this research is not that AI doesn't work — it's that most organisations under-budgeted for the human oversight it demands. Every automation shifts effort somewhere; the question is whether leaders planned for where it lands.

Efficiency gains from AI are rarely automatic — they are designed, and design has a cost that gets paid in governance, monitoring and human judgement long after go-live. Businesses that treat AI rollout as a one-time IT project, rather than an ongoing operating model with clear ownership of oversight and escalation, will keep rediscovering "botsitting" as a hidden line item. The fix isn't less AI; it's building the operational scaffolding — defined accountability, monitoring cadences, and a realistic view of staff time — before the tool goes live, not after the workload complaints start.