A cocked pistol on the road near Krushopek: the line between a police bulletin and a funeral was held by somebody else's nerve
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
06.08.2026
06.08.2026
07.08.2026
07.08.2026
06.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
06.08.2026
05.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
07.08.2026
09.03.2026
27.02.2026
19.02.2026
07.08.2026
06.08.2026
05.08.2026
23.04.2026
23.04.2026
12.04.2026
While Silicon Valley throws billions at startups promising that artificial intelligence will answer every question from every customer, a company from Athens has raised $67 million (around €62 million) on the opposite message: most of those questions don't require artificial intelligence at all.
The company is called Omilia and it has been automating phone calls and customer support since 2002. That isn't a small detail - that's twenty-four years in a business most of today's competitors discovered two years ago. The round is led by Expedition Growth Capital and is only the second in the company's history, after $20 million from Grafton Capital in 2020. Since then its annual recurring revenue has grown tenfold, to $60 million.
Chief executive Dimitris Vassos doesn't bother being diplomatic about the competition. "We'll use every available weapon to win the battle for customer service," he says. "Companies like Sierra, Decagon and the rest identify themselves as generative AI firms. Their only goal is to apply generative AI, and that limits them. You can have a bazooka, but if the enemy is close to you, you need a knife."
The argument is more concrete than it sounds. A large share of the calls coming into a contact centre are requests for basic information - account balance, order status, opening hours. That doesn't need a large language model that costs money on every question. Vassos argues this is exactly why Omilia has healthy unit economics both for itself and for its clients, and exactly why it never needed the big capital injections its competitors did.
"Over the next few years we'll see that the companies that survive are the ones offering real return on investment," Vassos says. "We don't mind that we're not as attractive as ElevenLabs and Sierra on LinkedIn right now. What interests us is stable growth and laying the foundations for a billion-dollar-revenue company in the next three years."
The client list keeps that statement standing: Capital One, Discover, RBC, DWP, PSEG. The newest focus is fast food chains that have introduced voice ordering - Taco Bell is already a client, with the technology deployed in over 1,000 outlets, and talks are under way with two more American chains. The company has around 500 employees and expects to reach 600 by the end of the year.
There's also one story Vassos rejects. Last year it was reported that a fault in Taco Bell's ordering system let a customer order 18,000 cups of water. Vassos says it never happened and that Omilia's logs show no such event. Whoever is right, the case shows what automation looks like when it goes wrong - and why a firm with twenty-four years in this field is careful about when to switch the machine on at all.
For a Balkan reader there's more here than another funding round. A Greek company, founded when "artificial intelligence" was an academic term, today bills American banks and runs technology in a thousand restaurants. The path was slow - two funding rounds in twenty-four years is almost impossibly slow by the standards of today's industry. The question that remains is whether that pace is a weakness or precisely the reason the company still exists.
The latest 10 news from this category
A company founded in Nigeria that financed cars for drivers with no access to credit is now worth 2.1 billion...
The company did not collapse - it has 480 million in revenue and 500,000 customers. It simply stopped being worth...
Instead of buying land next to the grid, the company puts the battery at the customer's house and then sells...
A company whose business rests on people driving their own cars is now building a platform where the human is...
There is more money for young founders than ever. Forgiveness for mistakes - none at all, and every wrong step...
Two studies mapped how fraud is manufactured inside tech firms. The conclusion isn't that founders are bad people, but that...
A startup that raised 29 million is selling for 200. The market is no longer growing on new users, but...
Pangram sells the assurance that it knows what a machine wrote. The technology works better than you'd expect - and...
Runlayer claims Rippling used the trial to build the same product. It's the trap facing anyone selling AI infrastructure to...
A lab less than a year old is putting 100 robots online to see how people want to talk to...
This site uses cookies - is that okay? Learn more
Be the first to know when Metla launches something new
Leave your email and we will write when there is a new guide or something new on Metla.