Revisión de la información de tu cuenta

Sat on hold for 22 minutes yesterday. Transferred twice. Never resolved. The company has an AI chatbot, apparently. It told me my issue was sorted. It wasn't.

Forrester published a workforce model this May that deserves more attention than it got. Take a contact centre with 1,000 customer service representatives. By 2030, according to Forrester's projection, that team could shrink to 40. Not 500. Forty. The other 960 aren't being laid off in a dramatic announcement. They just won't be replaced when they leave.

That distinction matters. Gartner's Eric Keller put it plainly: "headcount cuts are the exception, not the rule." Contact centre turnover already runs at 30 to 45 percent annually. Companies don't need to fire anyone. They just stop backfilling. The workforce contracts quietly, one resignation at a time, and the AI absorbs the volume.

It's already happening at scale. Microsoft reduced its support workforce from 50,000 to 40,000, a detail that got buried beneath the chip stories and product launches. Verizon's CEO told Bloomberg that AI is set to take over a large share of the company's customer service function — which landed as a business story rather than the labour story it actually is.

Forrester frames the 49 percent figure as a redefinition, not an elimination. Kate Leggett, Forrester VP, argues that "fewer people will answer routine questions and many more will direct, coach, and govern AI systems." That's a reasonable framing for the workers who make it through. For the ones who don't, the framing doesn't change much.

That's the true cost of replacing a single contact centre agent once you factor in lost productivity, not just recruiting and training. Insignia Resources puts the direct replacement figure at $10,000 to $20,000, but the total impact including productivity loss reaches $46,000 per agent. With industry turnover running at 40 to 45 percent annually, a 100-seat centre is effectively spending between $1.7 and $4.6 million a year on churn alone. That's before you buy a single AI licence.

The case for AI in contact centres keeps getting made on cost-per-call grounds. Voice AI costs $0.40 per call. Human agents cost $7 to $12. The maths looks obvious. But the stronger argument isn't the cost gap — it's what the cost gap is hiding.

Contact centres are already spending a fortune replacing people who burn out in under 15 months. Average agent tenure sits at 14 to 15 months industry-wide, and 87% of agents report high workplace stress. AI isn't replacing a stable workforce. It's filling the gap left by a workforce that was already leaving. That's a different conversation, and nobody is quite having it.

Hybrid AI-human models achieve 87% resolution rates with 8.7 out of 10 customer satisfaction, which beats full automation benchmarks across the board. The companies treating this as a binary choice — replace or don't replace — are asking the wrong question entirely.

The contact centre software market is growing at 21.9% CAGR through 2033 because most operations still run across 3.9 different platforms that don't share data. These three are worth your time right now.

• Cognigy Agent Assist — Surfaces relevant knowledge and suggested responses to human agents in real time, mid-conversation. It uses the same AI models that power autonomous agents, so the handoff between AI and human is coherent rather than a reset. Best for large, complex contact centre operations where voice quality and multilingual consistency matter.

• Kore.ai XO Platform — Builds, deploys, and manages AI agents across 35-plus channels and 100-plus languages. Strong governance and auditability framework makes it credible in regulated industries where you can't afford an AI that goes off-script.

• Intercom Fin — Resolves up to 50% of support volume against existing help centre content, with a pricing model tied to resolutions rather than seats. Useful forcing function: if Fin can't resolve it, you immediately know your documentation is the actual problem.

Thinking about that 14-month average tenure figure. Most contact centre agents leave before they've learned everything worth knowing. That's not an AI problem. It was already a management problem.