A Week Built Around the Right Idea 

Every September, thousands of contact centers pause to say thank you. International Contact Center Week has run since 2008 on one simple message: contact centers are about people. 

That premise is right, but it only holds if the day-to-day backs it up. When agents go back to fighting five disconnected systems, repeating the same answers all day, and drowning in backlog, the appreciation doesn't land.  

In this blogpost we break down four specific problems still holding contact centers back, from cost-center budgeting to rigid IVR menus, and shows where AI can close the gap versus where it's just noise.  

Problem #1: Still Treated as a Cost Center, not an investment 

Most organizations still budget and discuss the contact center the same way: cost per contact, average handle time, headcount, all numbers to minimize. 

Every improvement has to justify itself as a cost cut, because that's the only language the function is allowed to speak. Investment gets treated as expense, so it's the first thing trimmed when budgets tighten and the last thing modernized when they don't. 

What's different in the contact centers pulling ahead: they've changed the internal conversation. The contact center is where retention, upselling, and brand trust get decided, one interaction at a time. It gets budgeted like the revenue-protecting function it is, not the overhead line it's been filed under for decades. 

Problem #2: The First Step Is Still a Wall 

Picture a customer who just noticed a charge they don't recognize. They're already a little anxious, and they want one thing: to talk to someone who can explain it but instead they encounter the "Press 1 for billing, press 2 for support."  

They're not entirely sure which one covers "mystery charge," so they guess. Wrong department, wrong queue, start over, problem still unsolved. 

• It happens on voice, when a menu makes someone guess between "billing" and "support" for a problem that's neither. 

  
  

• The same thing happens on chat, when a scripted bot only understands a handful of pre-written options and none of them fit.  

  
  

• It happens on the web, when a support form makes someone pick from a dropdown that doesn't match what's wrong.  

  
  

Same failure, three channels: the customer ends up doing the classification work the system should have done, before anyone even looks at the actual situation. 

Why it matters: this is usually the first impression of the whole interaction. For someone who already has a problem, a fixed menu is friction stacked on frustration. 

Problem #3: "Digital-First" Leaves Some Customers Behind 

Contact centers are pushing more of the experience toward self-service, chat, and AI deflection. The assumption baked into most of these rollouts is that everyone is equally comfortable typing into a chatbot instead of talking to a person. 

A meaningful share of customers, often but not only older ones, find the "modernized" journey harder than the process it replaced. They end up more frustrated by the fix than by the original issue. An evolution that quietly makes things worse for part of the customer base is just automation with a blind spot. 

The business case: every customer pushed out of a journey they can't navigate is one bad experience away from calling a competitor, or telling their bank, insurer, or provider's social media why they switched. Investing in a well-supported human fallback it's protecting the revenue a purely digital rollout quietly puts at risk. 

Problem #4: Channels Don't Talk to Each Other 

A customer starts on chat, gets bounced to email, then calls in when neither worked. At each step, they explain the problem again from scratch, because the systems were never designed to share context. 

For the customer, this reads as not being listened to. For the agent picking up the call, it means starting cold on a problem that's already three touchpoints deep: no history, no context, just a frustrated customer repeating themselves for the third time. 

This is a solvable problem and the technology to unify context across channels has existed for years. What's usually missing is the decision to treat channels as one conversation instead of separate departments. 

What's at stake: every touchpoint a customer must repeat themselves is added handle time, an extra chance for the agent to say the wrong thing without the full picture, and one more reason resolution takes three contacts instead of one. Unifying context across channels isn't a nice-to-have integration project, it's the difference between paying for three contacts to solve one problem or paying for one. 

What Evolved Contact Centers are doing differently 

What separates an evolved contact center from the rest comes down to two things: where the automation sits, and what it frees people up to do. Headcount, tooling budget, and how much AI is in the stack matter less than how deliberately AI was placed. 

The centers that have made this shift tend to share the following three specific things. 

1. Replace rigid triage 

Modern conversational AI, built on generative models rather than fixed decision trees, works the same way across voice, chat, email, WhatsApp, and social messaging. It can: 

• Understand what a customer wants from how they naturally phrase it 

  
  

• Resolve simple requests outright (balance check, shipping update, subscription change) 

  
  

• Route everything else with real context attached, instead of a menu selection 

  
  

The goal is to make the human path easier and faster to reach, not to hide the phone number, the chat window, or the support inbox: a customer who does need a person gets there faster, and arrives with the system already knowing why they reached out. 

This matters most for the customers digital-first pushes tend to leave behind. An evolved center keeps the human path just as fast and well-supported as the automated one, on whichever channel someone reaches out through, instead of treating it as the option nobody wants you to take. 

2. Agents Working With AI 

Traditionally, an agent is the interface between systems that don't talk to each other: CRM, knowledge base, ticketing tool, billing platform, plus a compliance checklist running in their head. "Average handle time" ends up measuring how fast a human can do the job of five pieces of software. 

Agent-assist copilots close that gap. Instead of digging through a knowledge base mid-interaction, the agent gets the answer, the drafted response, or the compliance flag surfaced in real time. Attention stays on the customer, not the tooling. 

Some teams call this the "super agent" model: same person, same judgment but with dramatically less friction. It's also one of the more direct ways to act on "appreciate," appreciation that shows up in someone's actual workday, not a Slack message from HR. 

3. Backoffice Automation That Closes the Loop 

The least visible marker, and probably the most underrated, is what happens after the conversation ends. 

A huge share of agent frustration comes from what follows the conversation, not the conversation itself: 

• Updating three systems by hand 

  
  

• Re-entering data the customer already gave 

  
  

• Routing a case to another department and losing visibility into it 

  
  

Case and backoffice automation plugs directly into the CRM or ticketing system already in place, so updates, escalations, and multi-step processes happen without a human manually stitching systems together. The agent keeps the judgment calls; the software handles the plumbing. 

Behind the Scenes: Scripted to Agentic 

All three markers point at the same underlying change: automation is moving from scripted to agentic. 

• A scripted bot follows a decision tree built in advance. 

  
  

• An agentic AI system understands a goal, reasons through the steps to get there, and acts across multiple systems, with a human able to review, correct, and gradually extend how much autonomy it's trusted with. 

  
  

That "supervised to autonomous" progression is the honest answer to the fear underneath a lot of AI conversations in this industry: that automation is a slow-motion replacement plan for the workforce Contact Center Week exists to celebrate. 

In an evolved center, autonomy gets earned gradually, on the repetitive slice of work nobody wanted to do by hand, while judgment-heavy, relationship-heavy work stays exactly where it should: with a person. Channel doesn't matter either; customers don't experience your org chart, they just experience whether the problem gets solved. 

Seven Struggles, and Where AI Actually Helps 

Talking about "AI in the contact center" in the abstract doesn't help anyone. Here's the specific match between common struggles and what addresses them. 

1.Volume spikes that break the schedule 
A product outage, a billing error, a seasonal peak: contact volume across voice, chat, and email can double overnight. No staffing plan-built months ahead absorbs that cleanly.  
→ Conversational AI absorbs the spike itself, handling the flood of near-identical questions ("is this a known issue," "when will it be fixed") without a queue forming on any channel, so the human team stays focused on accounts that genuinely need judgment. 

2. Rigid first-contact triage is still the front door  
A fixed IVR menu, a scripted chatbot, a generic ticket form: all ask the customer to do the system's classification work, badly.  
→ Conversational AI replaces the fixed menu or script with something that understands the request in natural language on the first turn, on any channel, and hands off the rest with full context attached. 

3. Customers repeat themselves across channels  
A conversation that starts on chat and moves to a call starts over from zero, because the systems behind each channel don't share history.  
→ Unified, AI-Workflows carry the conversation history across channels, so whichever system or agent picks it up next already knows what's happened so far. 

4. New agents take months to become productive 
Knowledge bases are large, policies change constantly. A new hire is either slow and accurate or fast and wrong, with no good middle ground without support. 
→ Agent-assist surfaces the answer or compliance step in real time, compressing a six-month ramp-up because the agent isn't relying on memory alone from day one. 

5. The after-contact admin nobody sees 
Updating three systems, re-entering data, manually routing a case: a major, invisible driver of burnout. 
→ AI Workflows connects into the existing CRM or ticketing system, so follow-through happens automatically once the agent has made the judgment call. 

6. Inconsistent quality and compliance across a large team 
Two agents can handle the identical request two different ways, a real liability in regulated industries, not just a CX nuisance. 
→ The same assist layer flags the relevant compliance step for that specific interaction type, raising the floor for the whole team instead of relying on training alone. 

7. "Digital-first" leaves some customers stranded 
Self-service and chat deflection assume a comfort level with digital tools a meaningful share of customers, often but not only older ones, don't have. 
→ The fix is automation that recognizes when someone needs a person quickly and hands them off with context already gathered, protecting the human path for the customers who need it most. 

The common thread: the AI that moves the needle removes a specific, named piece of friction sitting between the customer and the person meant to help them, rather than removing the person. 

What This Looks Like During Contact Center Week 

A useful filter for anything planned for September 10–17: does it change anything about Monday, September 21st? 

A few concrete moves worth considering: 

• Publish the deflection numbers to the floor, not just to leadership. If conversational AI resolved a meaningful share of routine contacts last month, agents should hear "fewer repetitive contacts reaching you," not see it buried in a QBR deck. 

  
  

• Ask agents what the assistant tooling gets wrong. Agent-assist systems improve fastest when the people using them daily are the primary feedback source. 

  
  

• Show the after-contact workload, out loud. Most leadership teams underestimate how much unpaid cognitive labor sits in "wrap-up" time. Naming it, and showing what's being automated out of it, is respect a gift card can't match. 

  
  

• Talk about the trajectory, not just the current state. Agents trust AI more when they understand it's expanding their capability. Be explicit about what's automated, why, and what remains human. 

Eighteen years in, International Contact Center Week still runs on a simple, correct idea: this industry is powered by people, and they deserve to be seen. 

The centers that get the most out of this year's theme won't have the biggest banner. They'll have a concrete answer when an agent asks, "what changed for me this year?", fewer repetitive contacts reaching the queue, less time lost switching systems, less unpaid cleanup after every interaction, and more of the day spent on the part of the job that actually required a human being. 

That's an evolved contact center: one that used AI to remove everything standing between its people and the work they were hired to do. And this year, it's also one with a real story to tell while the rest of the industry is telling theirs. 

At Automaise, we work with enterprise contact centres across Europe to build that story. From Conversational AI that handles volume spikes to Agent-Assist that closes the knowledge gap and AI Workflows that connect every channel into one conversation, we help operations move from where they are to where they need to be.

Book a CX & AI Advisory Session and find out what that looks like for yours.