The pressure to automate CX has never been higher.
Boardrooms want cost control. Customers expect instant resolution. AI agents are getting more capable by the quarter.
But in our recent webinar, What CX Orgs Must Get Right in 2026: Building a Human + AI Workforce, AI expert Pascal Bornet challenged a core assumption driving many AI programs today:
Automation alone is not a strategy.
Yes, AI can reduce cost.
Yes, 60–70% of CX volume can likely be automated.
But the real competitive advantage won’t come from how much you automate. It will come from how well you design humans and AI to work together.
Here are the five most actionable takeaways from the conversation.
1. Stop Asking “What Can We Automate?”
Start asking:
“Where does the right balance between humans and AI create the most value?”
Pascal outlined three distinct categories of work in CX:
- Automated (AI-led, human-supervised)
High-volume, rule-based, low-complexity work, like password resets, order status, and appointment scheduling. - Augmented (Human-led, AI-assisted)
Complex but structured work, like escalations, complaint handling, and personalized retention offers. AI provides context, sentiment, and next-best actions. The human decides and executes. - Supported (Human-led, AI-enhanced)
High-value, brand-defining interactions, such as sensitive complaints, relationship-saving moments, and strategic CX improvements. AI removes the administrative burden, and humans create the value.
The mistake that most organizations make is treating everything as a candidate for automation.
The 60–70% of volume that can be automated drives efficiency. But, as Cresta’s VP of Product Marketing, Devon Mychal, pointed out, the remaining 30% often drives 70% of loyalty and revenue, and that’s where margin lives.
2. Some Conversations Shouldn’t Be Happening at All
One of the sharpest insights from the session was the advice to, before deciding whether to automate a conversation, ask:
Why is this conversation happening in the first place?
Are customers calling because billing is confusing? Order tracking is broken? Policies are inconsistent?
AI will scale whatever you give it — including broken processes. Fix root causes upstream, and you reduce volume permanently. Automation without process improvement just accelerates chaos.
This is where conversation intelligence becomes critical: surfacing the patterns, friction points, and repeat failure drivers hidden across thousands of interactions so you can fix what’s causing the contact in the first place — not just resolve it faster.
3. The Future CX Agent Isn’t “Faster.” They’re Different.
If AI absorbs routine volume, what happens to human agents?
They don’t disappear. They specialize. Pascal walked us through three core competencies every CX professional must develop to keep pace with AI innovation in the years to come:
1. Change-Ready
The tech stack will evolve every 6–12 months. Workflows will shift, AI capabilities will expand. Adaptability isn’t optional — it becomes a core professional skill.
2. AI-Ready
Agents must:
- Interpret AI outputs critically
- Prompt effectively
- Override when necessary
Blind trust in AI is as dangerous as rejecting it.
3. Human-Ready (The Most Important)
These are the “Humics” — uniquely human capabilities AI cannot replicate:
- Creativity — finding solutions that aren’t in the playbook
- Critical thinking — knowing when the AI suggestion is technically right but contextually wrong
- Human connection — empathy in high-stakes moments
These aren’t soft skills; they directly impact retention, revenue, and brand equity. If AI handles routine work and humans never practice these critical skills, they atrophy. Hybrid workforce design must deliberately protect and grow them.
But competencies alone aren’t enough. New roles must emerge to operationalize a hybrid model.
Pascal described the rise of the AI Orchestrator, a role responsible for allocating work between humans and AI, monitoring system performance, and managing handoff quality.
We also anticipate the evolution of this role, formalized in Cresta’s Agent Operations Center — a unified command center for managing, optimizing, and governing both human and AI agents in one system. In a hybrid workforce, orchestration between humans and AI isn’t a side responsibility. It’s a core operational function.
4. Governance Gets Harder — and More Important
Two key governance realities stood out:
Accountability
If an AI agent harms a customer relationship, who owns it? The answer must be defined before deployment. And accountability must always fall to a human.
QA Must Evolve
Random sampling worked when humans handled 40 contacts per day. AI handles thousands. You need statistical process control, not spot checks.
5. Measure What AI Creates — Not Just What It Removes
If you only measure metrics like containment rate, cost per contact, and handle time, you’re only tracking reduction.
Hybrid workforce measurement must expand to include metrics like:
- Resolution quality
- Customer trust
- Agent confidence
- Long-term revenue impact
- Loyalty and lifetime value
AI that “resolves” 80% of contacts but damages brand perception in the remaining 20% is not a success story.
To go deeper on the frameworks Pascal shared — including the four practical filters for deciding where AI should lead, assist, or stay out entirely — you can watch the full webinar replay here..
And if you’re ready to move from theory to execution, see how Cresta brings this hybrid workforce model to life. Cresta’s omnichannel AI Agent doesn’t just automate conversations; it augments human agents in real time, orchestrates seamless handoffs, and gives leaders the visibility and control needed to manage both human and AI performance in one unified system.
Schedule a personalized live demo now to see how you can operationalize AI safely, strategically, and at scale
