The Productivity Promise Is Running Into a Management Wall

When businesses invest in AI for customer service, the pitch is almost always the same: reduce costs, speed up response times, free your people for higher-value work. But new research highlighted by Freshworks — and corroborated by a growing body of 2026 CX studies — is surfacing an inconvenient truth: AI, deployed without discipline, is creating as much operational drag as it eliminates.

According to research covered by CX Today, AI complexity is measurably increasing the workloads of IT teams across U.K. businesses, draining budgets that were supposed to be freed up, and pulling technical staff away from the strategic initiatives that were meant to justify the AI investment in the first place. A companion concept gaining traction in industry research is "botsitting" — the quiet but very real phenomenon of skilled employees spending significant portions of their working day monitoring, correcting, and babysitting AI systems that are not yet reliable enough to operate unsupervised. Governance gaps compound the problem: many organisations deployed AI tools faster than they built the oversight frameworks to manage them responsibly.

This is not a niche IT problem. For CX and operations leaders, it lands directly on the contact centre floor.

What This Actually Looks Like Inside a Customer Operations Team

Imagine a contact centre that has deployed a conversational AI to handle first-line customer queries. On paper, containment rates look promising. But behind the scenes, a cluster of agents and team leads are spending hours each week reviewing transcripts for errors, manually escalating cases the bot mishandled, updating intent libraries, filing exceptions, and fielding complaints from customers who felt poorly served by an automated interaction. The AI did not replace that work — it transformed it into a less visible, less valued, and deeply frustrating form of overhead.

This is the governance gap in action. The bot is live, but nobody budgeted for the ongoing human effort required to keep it performing at an acceptable standard. Senior agents who should be handling complex, high-value cases are instead doing quality control on a system that was supposed to make their lives easier. Morale erodes. Attrition risk rises. And the ROI calculation that justified the AI deployment starts to look a lot less convincing.

Multiply this across multiple channels — chat, voice, email, social — and you can see how AI complexity compounds quickly at scale.

Why Hybrid Intelligence Is the Operationally Mature Response

The answer is not to abandon AI. The productivity gains are real, and the competitive pressure to deploy AI in customer operations is not going away. But the research is a clear signal that the "set it and forget it" model of AI deployment is a liability, not a strategy.

The organisations navigating this most effectively are those that have stopped thinking about AI and human talent as an either/or choice and started designing them as a single, integrated operational system. In that model, AI handles what it is genuinely good at — high-volume, low-complexity, well-defined interactions — while skilled human agents manage nuance, exceptions, emotionally sensitive cases, and crucially, the ongoing quality oversight of AI performance itself. That oversight role is not an afterthought. It is a defined function with clear accountability, built into resourcing plans from day one.

This is precisely where multilingual, premium human talent — embedded within an AI-enabled operating model — delivers disproportionate value. Experienced agents who understand both the customer journey and the behaviour of AI systems are the connective tissue that keeps the whole operation reliable. They catch what the bot misses, they escalate with context, and they feed the continuous improvement loop that stops AI quality from drifting over time.

The Operational Takeaway for CX Leaders

If your organisation has deployed AI in customer service without explicitly budgeting for human oversight capacity, governance processes, and regular performance audits, the Freshworks research is worth reading as a diagnostic, not just a news item. The complexity cost is already accumulating — it is just not yet showing up on your dashboard.

The smartest CX operations in 2025 are not the ones with the most AI. They are the ones where AI and human talent are designed to cover each other's blind spots. That balance is not a compromise — it is a competitive advantage.