Dean Koenning, founder of VOICO, joins Sunny Ray to discuss building GDPR compliant, EU hosted voice agent infrastructure for European businesses. Koenning traces his path from door to door sales at HelloFresh, through a head of sales and CSO role at a German OEM, to founding VOICO while still employed full time, a transition that took roughly eight to nine months before he went all in. He explains how VOICO brings AI agents to the telephone line, letting businesses like car dealerships answer calls, book appointments, connect to CRM and DMS systems, and even upsell customers, using car dealerships as VOICO's strongest use case. He shares how VOICO landed early customers including Kia by first building agents on Silicon Valley infrastructure, then rebuilding for GDPR compliance. Koenning also discusses building in public on YouTube, closing a friends and family funding round at Bits and Pretzels in Munich, the limits of what AI voice still cannot do, and his prediction that voice agents will fully replace IVR systems and voicemail within a couple of years.
A door to door HelloFresh salesman turned founder building GDPR compliant AI voice agents that answer Europe's business phones.
What is VOICO, and why should the world care?
VOICO is GDPR compliant, EU hosted voice agent infrastructure. We bring AI to the phone line for European businesses, turning agents into real employees who can answer calls 24/7, make outbound calls, and act like a human colleague. It's not only about voice. The agent is also wired into a company's software so it can take real action after the call, helping teams grow without hiring more people.
You started out doing door to door sales for HelloFresh. What did that teach you that no startup advice ever could?
Every founder eventually has to do real sales, whether that's door to door or cold calling. I was lucky enough to do it more than ten years ago with HelloFresh, and it taught me the raw first interaction with a client: going out on the streets, picking up the phone, and finding customers yourself. It's not easy, but it's the most crucial part of building any startup.
Walk me through a car dealership using VOICO. If a phone rings there, what happens in the next 90 seconds?
Car dealerships are our best ICP. Most callers barely notice they're talking to an agent instead of a person. The agent picks up, speaks the dealership's language, and is trained on their systems, software skills, and DMS or calendar setup. It can book a service appointment, connect it to the calendar, and then upsell, asking if you want your oil or tires checked too. The results have been impressive.
How did VOICO land paying customers before the MVP even launched?
Probably my door to door background made me pretty direct about sales, so I acquired pilot customers before the MVP existed. We started as an AI agency in Germany implementing voice agents and chatbots, and I noticed there was no GDPR compliant option in Europe. I built the first agents on Silicon Valley platforms to prove product market fit, then rebuilt everything to be compliant for the European market. Within our first year we landed Kia as a client.
What's the one question a German buyer asks that kills most American AI vendors on the spot?
German buyers care a lot about data protection, and GDPR compliance is necessary for basically every company here. Once you hit around ten employees, or a certain call volume, you have to address these regulations. People also like working with companies based in their own country. We still let clients use foreign models when they want the absolute best performance and lowest latency, but compliance is usually the deciding factor.
What did the first version of VOICO get wrong?
Honestly, the real question is what did it get right. It took eight months to build the MVP, and I assumed that once it launched we could just scale. That's not what happened. The actual work started after launch, with customer feedback and new requirements coming in daily. People assume you can spin up your own voice agent platform easily with today's tools, but building something that truly works is a never ending process.
What can AI voice still not do, and do you tell customers that upfront?
Absolutely. You have to make customers conscious of the agent's real capabilities, or they'll assume AI can do everything. We can typically get an agent to about 80 percent effectiveness, never 100, because some hallucination is always possible. The biggest limit is legacy software with no API to connect to. We also won't build agents to cold call someone's old lead lists, since that runs into real regulatory issues, even though it's technically possible.
If VOICO wins, what will calling a business in Europe feel like in five years?
It's hard to say exactly what the world looks like in five years, but voice in general will become far more natural, not just on the phone but everywhere we interact with software. On calls, you won't tell a human from an agent, and agents will often be more capable since they know a company's whole knowledge base instantly. I don't think it will even take five years for voicemail and IVR menus to disappear completely.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview