The AI industry frequently promotes the largest and newest model as the answer to every business problem.
Omilia’s chief executive has a simpler message:
Sometimes, using a giant generative AI model is like bringing a bazooka to a knife fight.
The Athens-based customer-support technology company has raised $67 million to expand its self-learning agents across telephone calls, chat, and messaging.
Omilia has worked on voice and customer-service automation since 2002. Its chief executive, Dimitris Vassos, argues that many support requests—such as checking an account balance—do not require a powerful language model at all. �
TechCrunch
Why More AI Is Not Always Better
Businesses often assume every customer interaction should be sent to the smartest model available.
That approach can be wasteful.
Simple questions may be handled through:
Structured databases
Traditional automation
Rules-based systems
Small classification models
Pre-approved responses
A frontier language model becomes valuable when the customer’s request is unclear, emotional, complicated, or requires reasoning across several systems.
Using expensive generative AI for a basic balance check increases cost without necessarily improving the answer.
Omilia’s Multi-Tool Strategy
Vassos says contact centres need several tools rather than one universal system.
The company is developing agents capable of operating across multiple communication channels while selecting the appropriate technology for each type of request. �
TechCrunch
This is practical AI orchestration:
Routine request: low-cost automation
Complex language problem: generative AI
Sensitive complaint: trained human
Fraud risk: specialised security system
High-impact decision: human approval
The intelligence lies partly in knowing when not to use the most powerful model.
The Money Behind the Strategy
Omilia says its annual recurring revenue has increased tenfold since its previous funding round, reaching approximately $60 million.
The company argues that strong unit economics—making each customer interaction financially sustainable—will eventually matters more than appearing fashionable in online AI conversations. �
TechCrunch
That is an important signal for the wider industry.
Investors have funded many companies promising fully automated customer support.
The next phase will test whether those systems genuinely reduce costs without frustrating customers or creating expensive mistakes.
The Human Problem
Customer service is not merely information retrieval.
A customer may technically ask about a delayed payment but emotionally need reassurance that the problem is being taken seriously.
AI can recognise language patterns, but businesses must remain careful when handling:
Vulnerable customers
Bereavement
Debt
Medical issues
Fraud
Complaints
Account closure
Legal disputes
Automation should make human assistance easier to reach—not trap customers inside an endless synthetic conversation.
Why This Story Is Viral
The CEO’s metaphor punctures a central piece of AI hype.
The smartest tool is not automatically the best tool.
A business that pays frontier-model prices for every minor request may discover that its AI assistant is technically brilliant and financially ridiculous.
That is how the Scaler Bill arrives wearing a customer-service headset. 🤣
Mary Chuks’ Perspective
This is exactly the philosophy behind multi-AI orchestration.
Use the right intelligence for the right job.
Businesses do not need one model attempting to become a researcher, receptionist, accountant, therapist, and technical engineer simultaneously.
They need a system that routes tasks sensibly.
The strongest AI product may not generate the most impressive answer.
It may deliver the correct outcome at the lowest safe cost.
Conclusion
Omilia’s funding reflects a growing business reality:
The AI winners will not merely automate more work.
They will understand which work deserves which tool.
Sometimes, you need the frontier model.
Sometimes, you need a database lookup.
Sometimes, the customer needs a human being who can simply say:
“I understand. Let me fix this.”