Sensors are the unsung heroes of modern life. They keep us safe, alerting us to everything from fire, flood and house intrusions to low blood sugar levels and slow punctures. But alerts without action are useless, and this is where monitoring centres step in.
The challenge these businesses face is scale. As sensors become even more widely deployed, the number of alerts keeps rising, and centres are swamped by the sheer volume of calls. Until recently, the only solution was to hire more staff, which added to costs and didn’t solve the problem of sudden or seasonal spikes in calls.
Enter AI voice technology. With more sophisticated text-to-speech and speech-to-text capabilities - and the technical know-how of software engineer Mark Harkin - suddenly there’s another option: an AI voice agent that can do the heavy lifting on the thousands of repetitive, low-risk alerts that tie up human call-takers for most of their working day.
Vox Talk AI, founded earlier this year, grew out of Harkin’s long-standing fascination with the possibilities of AI and large language models. Harkin already enjoyed building them as a hobby. “When the latest breakthroughs came, I was blown away by how natural and capable it had become,” he says.
“In the very near future, we will interact with AI voice daily, and the adoption curve will only compound over the next couple of years. I wanted to be part of that shift. I began researching industries where voice AI could have the most impact, and identified the security and alarm monitoring sectors as having significant potential, as they rely heavily on human operators.”
[ Irish fintech Harbour AI takes aim at shipping’s late-payment problemOpens in new window ]
To test the idea, Harkin spoke with more than 50 response centres across Ireland, the UK, Europe and North America. As the conversation progressed, it became clear that an AI voice could ease a major pain point.
“We are targeting the global security and monitoring market, which is valued at over €70 billion and is constantly being challenged by the pressures of managing scale, costs and compliance,” Harkin says.
Dublin Bus CEO on recruitment challenges, going electric, and stamping out anti-social behaviour
“Unlike human operators, our AI agents can handle hundreds of concurrent calls, eliminating wait times and telephone queues. In addition, we support over 30 languages, thereby giving security companies a unique opportunity to scale internationally. Our platform handles industry-specific workflows with built-in compliance and replaces outdated IVR (interactive voice response) systems with natural, human-like conversations.”
A major boost for Vox Talk came when it integrated directly with Sentinel, a widely used alarm-response software developed by Monitor Computer Systems in York.
“Sentinel is a leading ARC [alarm response centre] software provider with hundreds of global clients, and Vox Talk is now its designated AI voice provider,” Harkin says.
This is a major coup for a fledgling start-up, so how exactly did Harkin pull that off? “I quickly realised that for my software to work, it had to integrate into existing technology stacks,” he says. “I was talking to potential customers - already using Sentinel - and they did the initial introductions for me. I then set about building a relationship with Monitor, who could see where Vox Talk could add value.
“I worked with their development team for a number of months, and we subsequently agreed on a strategic partnership with a non-compete clause,” Harkin says. “We launched commercially in July at a webinar with 50 of their global clients, and off the back of this, I was inundated with companies wanting to trial the software.”
Harkin notes that property security alarms account for only a small part of the sector. Monitoring centres have a very broad remit and can be dealing with anything from medical issues to electrical fault alerts at data centres to temperature drops in frozen food cabinets in big supermarkets. Lone workers, such as health professionals visiting people at home or engineers working at remote sites, also represent a large use case cohort.
“In North America, long distance lorry drivers have to check in every hour as required by their insurers. If they fail to do so, a human has to call them,” Harkin says. “Our AI agent can handle this type of routine call in multiple languages, leaving the human operators free to deal with more serious incidents.”
Vox Talk’s software can listen and speak. The next step is to add computer vision. “We are working on adding ‘eyes’ to our system so it can see as well as hear and talk,” he says.
Harkin self-funded Vox Talk to the tune of about €50,000 and is currently in the process of raising €800,000. The business had its first paying clients within six weeks of launch and operates a tiered pay-per-minute pricing plan, which is billed monthly.
Harkin is currently on the New Frontiers programme at TU Tallaght, and previously took part in the National Digital Research Centre’s founder accelerator at Dogpatch Labs, which aims to match those interested in starting a business with a co-founder. Harkin went through the process, but ultimately decided he was happy to go it alone.
“It was an avenue worth exploring, but with previous start-up experience and a career spent in senior management in the corporate sector, I was comfortable with being a sole founder,” he says.
“The combination of mentoring, grant funding, and accelerator programmes I’ve availed of helped me move quickly from idea to pilot deployments. However, one improvement that would really benefit start-ups is faster access to follow-on funding, to bridge the gap between pilot traction and seed investment.”
