Omilia’s $67M Series B: Why Its CEO Says Most Contact Centers Are Using AI All Wrong

Key products include Voice Agents, Chat Agents, TalkGuard (voice biometrics for authentication), a CSR CoPilot, and Workforce AI tools .

Key Differentiators

• Voice-first, self-learning architecture – The platform was built from the ground up for voice, claims sub-45ms latency with its proprietary Lexis text-to-speech engine, and can handle over 50,000 concurrent voice calls .
• Hybrid AI ("right-sized" approach) – It uses deterministic tools for simple, structured tasks and generative AI only where it adds value, rather than deploying an LLM on every query .
• Enterprise security and scale – The platform supports a Tier 1 U.S. bank handling over 1 million calls per day, with multi-layered anti-fraud mechanisms .
• No data prep bottleneck – The self-learning system connects directly to live operations and learns from scratch without requiring curated datasets .
• Analyst recognition – Named a Leader in the Forrester Wave™ Conversational AI Platforms for Customer Service, Q2 2026 and recognized in the IDC MarketScape .

The CEO's Argument: 'Bazooka vs. Knife'

Vassos makes a sharp, pragmatic case against deploying generative AI on every contact center interaction . His argument rests on a vivid analogy: using a large, expensive LLM for every request is like using a bazooka when a knife will do . Routine tasks — checking an account balance, payment status, or order details — do not require a frontier model .

"Companies like Sierra, Decagon, etc. identify as generative AI companies. Their sole purpose is to deploy generative AI and limit themselves," Vassos told TechCrunch. "You may have a bazooka, but if your enemy is near you, you need a knife. This is the reality of the contact center, where you need multiple tools."

A large proportion of incoming support calls involve basic, deterministic information retrieval (e.g., "What is my balance?"). These can be handled more cheaply and reliably by task-specific automation . The core thesis is using the least expensive tool that reliably resolves the query, with Omilia's hybrid approach claiming to beat an all-LLM stack on cost per correctly contained call .

Recent Funding and Growth Metrics

Omilia has only conducted two external equity rounds in its history. The Series B of $67 million (€58.1 million) was led by Expedition Growth Capital, a transatlantic software and AI specialist . The previous Series A of $20 million came from Grafton Capital in 2020 .

Crucially, the company grew its live annual recurring revenue from an undisclosed base at Series A to over $60 million without raising any additional equity capital in between . This capital efficiency — growing 10x+ ARR with only one early equity round — is atypical among AI companies burning cash on token costs .

Additional growth benchmarks:

• Supports more than 200 enterprise deployments globally, including at least 27 production-grade contact centers
• Automates more than 90% of customer interactions for its clients
• Handles over 5 billion interactions per year on the Omilia Cloud Platform

Clientele

Omilia serves large, heavily regulated enterprises. Confirmed details include:

• A Tier 1 U.S. bank handling over 1 million calls per day
• Multiple financial services and insurance firms across North America and Europe
• At least 27 production-grade contact centers globally, per the company's Microsoft Marketplace listing

Industries served include banking, insurance, utilities, QSR, telecoms, and public services .

Expansion Plans

The Series B capital is being used to:

• Open a U.S. office and build North American go-to-market presence
• Expand the go-to-market team and accelerate technology scaling
• Deepen DACH and Benelux presence through a partnership with valantic, making the platform available to valantic's 500+ enterprise clients
• Launch a UK-wide service in partnership with Bell Integration
• Position the company toward a billion-dollar revenue run-rate within three years

The Bottom Line

Omilia represents a counterpoint to the prevailing narrative in customer support AI: that generative AI should handle everything. By combining deterministic automation for routine tasks with generative AI for complex interactions, Vassos argues that contact centers can achieve better unit economics, lower latency, and stronger compliance. The company's 10x ARR growth and $67 million Series B suggest that enterprise buyers agree.