Home > Columns > Executive Interviews

Parloa Executive Interview

Click image below to read the ebook

In this interview, Sheri Greenhaus, Managing Partner at CrmXchange, speaks with Malte Kosub, CEO and Co-Founder of Parloa, about revolutionizing customer service with Generative AI. Malte explains how Parloa’s AI-first approach surpasses traditional chatbots, delivering personalized and empathetic experiences that enhance customer satisfaction. The discussion covers how businesses can start small, scale efficiently, and ensure AI reliability through simulations and evaluations. Discover how Parloa is reshaping customer engagement and setting a new standard for enterprise-level AI-driven interactions.

Sheri Greenhaus: Tell me a little bit about why and how you started Parola.

Malte Kosub: In 2016–2017, my co-founder and I met while exploring different technologies. Stefan, my co-founder, was already very focused on voice technologies because, in 2017, speech-to-text technology was skyrocketing in accuracy. It was reaching the capability of humans in accurately understanding audio, which was a breakthrough.

That was the first sign that speech technology and AI were advancing to a point where they could outperform humans. We realized this could revolutionize how companies interact with customers. We started by building a voice-focused agency to help companies leverage AI in customer communication.

Out of that agency, we saw a significant gap in enterprise tooling for developing and deploying AI into contact centers. So, in 2018, we started building Parola. Since then, the company has grown significantly.

Sheri Greenhaus: Where did the interest in contact centers or customer service come from?

Malte Kosub: After observing the breakthrough in technology, we asked ourselves where it would have the biggest short-term impact. We thought about our own frustrating customer support experiences and realized that enterprises also struggled to deliver great customer experiences.

Moreover, agents found their roles repetitive and frustrating. This showed us that the contact center market was broken. We saw the rapidly advancing technology as a solution to this problem, and that is why we decided to focus on customer support and contact centers.

Sheri Greenhaus: Are most of your customers currently in Europe?

Malte Kosub: Yes, most of our customers are in Europe because we only started in the U.S. market in 2023. However, the U.S. has become our fastest-growing market. We’ve already gained traction with several Fortune 500 companies here.

Sheri Greenhaus: If we were in an elevator and you had ten floors to explain what Parloa does, what would you say?

Malte Kosub: We believe in a world where every interaction with a company can be as easy as talking to a friend. Imagine calling an airline, and within one second, an AI agent answers, saying:

"Hi Sheri, how are you? Have you decided if you’d like to upgrade to business class? No worries if not. Call back anytime."

If you call back in five minutes, the same AI agent picks up instantly and continues the conversation seamlessly. You could also send a WhatsApp message to this AI agent to upgrade your flight. Alternatively, the AI might proactively call you later, saying:

"Hey Sheri, only three seats are left for that upgrade. Would you like one?"

We aim to create a world where every customer has a personal, dedicated AI agent that assists them throughout their journey. Parloa helps companies deliver these personalized AI interactions.

Sheri Greenhaus: Are the AI agents equipped for both voice and chat? Will customers know they are interacting with AI?

Malte Kosub: Yes, the AI agents work across both voice and chat channels. Our recommendation to customers is to be transparent and disclose that it’s a virtual assistant or agent. However, the platform allows businesses to configure this based on their preferences, so it’s ultimately up to them.

Sheri Greenhaus: Let’s say I’m a bank and want specific types of voices—like a southern accent or a faster-paced tone. Is that possible?

Malte Kosub: Yes, companies can choose different voices based on regions and even specific customers. While this is already possible to some extent, we believe future advancements in end-to-end audio models will make it even more seamless.

Currently, speech-to-text and text-to-speech are common. In the future, end-to-end audio models will enable real-time adjustments like changing accents or languages on the fly. These models aren’t enterprise-grade yet, but they’re rapidly improving and could be ready within the next 12–18 months.

Sheri Greenhaus: If I am interacting with an airline, can I get the same AI agent across different channels, like calls and text?

Malte Kosub: Yes, the same AI agent can handle interactions across multiple channels, whether it’s a call, text, or chat. This ensures a consistent and seamless customer experience.

This is your personal dedicated AI, always available to you—whether on the phone, WhatsApp, or the homepage. This AI agent knows your entire history with the airline. You don’t need to specify which flight you’re talking about. You can simply say, “Hey, my flight next week,” and the AI agent will know what you mean. If you’re using the airline app or WhatsApp, you’re already identified, so the agent instantly understands your context.

Sheri Greenhaus: So how does it work? Is it accessing APIs or databases? How does it know my history?

Malte Kosub: We’re connected to an enterprise's backend systems, like Salesforce, SAP, or any CRM or ERP they use. The customer is identified either through a unique identifier on WhatsApp or the homepage. On phone calls, we might need to ask one or two questions for verification. Using APIs, we gather information like past interactions, flight bookings, or previous inquiries. This enables the AI agent to have a context-aware conversation based on customer data.

Sheri Greenhaus: How complex can these conversations get? Simple tasks like upgrading a flight or fixing an order are straightforward. What about more challenging requests?

Malte Kosub: The AI agent can handle much more complex interactions than traditional chatbots. For example, you could say:

“I need to be in Midtown, New York, in two days at 2 PM. Find me a flight.”

The AI agent can calculate the time needed for traffic, consider different airports, and recommend flights that land two or three hours prior. Or you could say:

“I want to go somewhere sunny this weekend from New York. What do you recommend?”

The agent might suggest Miami, or another sunny destination based on your preferences. This level of complexity is far beyond the capabilities of older chatbot technologies.

Sheri Greenhaus: Is this where you see your platform differentiating itself?

Malte Kosub: Our differentiation is on several levels:

1. Generative AI-First Platform:
   Most conversational AI tools today are built on workflow- and rule-based engines infused with a bit of generative AI. In contrast, we built our platform from scratch as a generative AI-first product. Every decision, reasoning, and action is driven by large language models. This gives us unmatched flexibility and quality.
2. Enterprise Focus:
   We cater specifically to large enterprises. Currently, we work with two of the ten biggest retailers, three of the ten largest insurance companies, and even a Fortune 10 client.
3. Business Enablement:
   We empower business units to deploy, test, and scale AI agents effectively. For example, we offer simulation and evaluation tools that allow teams to simulate tens of thousands of conversations before launching an AI agent. This ensures performance is optimized before going live.

To sum it up: we’re generative AI-first, enterprise-focused, and heavily invested in enabling business teams to succeed at scale.

Sheri Greenhaus: When I use a chat agent, it’s often a terrible experience. I can’t get the help I need. At the dog park, people always complain to me about bad customer service. Why do companies settle for solutions that damage their brand instead of improving it?

Malte Kosub: We are now at a turning point where AI can genuinely enhance customer experiences. For example, one of our clients—a health insurance company—reported that 60% of customers said their relationship with the brand improved after interacting with the AI agent.

Why is this happening now? Because today’s AI can finally deliver the value we’ve been talking about for the last decade. It’s always available, empathetic, and friendly. It works faster than humans, instantly accessing and summarizing data from APIs to provide accurate responses. The AI agent also remembers your history, ensuring consistent and personalized interactions.

The technology is finally here, but enterprises need time to adapt and integrate it. Over the next few years, I believe we’ll see significant changes as more companies adopt AI that truly enhances customer experience.

Sheri Greenhaus: Do most of your customers already have chat agents, or is this a brand-new solution for them?

Malte Kosub: Most of them already had chatbots in place. They weren’t necessarily saying, “This isn’t working,” but rather, “It’s just okay.” These chatbots made customer service cheaper than using humans, but they didn’t enhance the brand. Now, with our solution, they’re seeing it actually creates a better customer experience.

Sheri Greenhaus: It’s interesting—we recently did surveys with DMG Consulting LLC. The survey we’re presenting in December focuses on AI and GenAI. One question we asked was, “What do you want to achieve with AI?” Years ago, the top answer was cost reduction. Now, 60% said they want a better customer experience, with cost reduction falling to third. It’s a shift because companies are realizing, especially with social media, how quickly bad experiences can damage their brand.

Malte Kosub: I completely agree. It’s a mindset shift. Now that the technology can genuinely improve customer experience, businesses are recognizing its value. Customer experience has become the top priority.

Sheri Greenhaus: At conferences, when you discuss your solution, what’s the reaction from attendees?

Malte Kosub: People get excited. They want to try it, test it, and explore the nuances of what it can do. It’s not just 10% better than a chatbot—it’s ten times better. That’s what generates so much enthusiasm.

Sheri Greenhaus: Here’s an example: I was using Microsoft’s Copilot to discuss a book, and I ended up arguing with the AI. It was so engaging that I stayed on far longer than I intended. Do you see scenarios where your AI could expand into deeper conversations, like for older demographics who may feel isolated and want to talk beyond their immediate needs?

Malte Kosub: It’s an interesting and even philosophical question. Companies must decide where their value creation stops. For example, an airline’s AI agent might focus strictly on flights and travel-related topics, while another company might allow their AI to engage in broader, conversational interactions. It depends on the brand’s strategy and the value they want to deliver. AI has the potential to do so much more than handle simple tasks—it could create entirely new dimensions of customer interaction.

Sheri Greenhaus: If a company wants to implement your solution, it seems complex. How do they get started?

Malte Kosub: It’s actually simpler than it sounds. We always recommend starting small. Look at high-volume, easily solvable use cases in first-level customer support—areas with clear APIs and knowledge bases. Start with those.

For example, many companies still use clunky touch-tone IVRs or keyword-based systems. A great entry point is a concierge AI agent that welcomes customers, identifies their needs, and routes them correctly. It’s a straightforward use case that can go live in a few weeks, improving the customer experience immediately. From there, you can expand into authentication, FAQs, or other processes. The key is to start and develop the skill set to scale.

Sheri Greenhaus: Are companies, like the Fortune 500 or insurance companies, starting small because they see the potential to grow?

Malte Kosub: Yes, there are typically two starting points.

1. IVR Replacement: Companies replace outdated IVR systems with an AI-powered solution to create a better initial experience.
2. Specific Use Cases: Companies tackle one process, like an API-driven task, to prove the value of AI in that area.

Both approaches have their merits, and companies often choose based on their immediate priorities.

Sheri Greenhaus: You mentioned simulations. What role do they play?

Malte Kosub: Simulations allow companies to test their AI agents at scale before going live. They can simulate thousands of conversations to evaluate performance, fine-tune responses, and ensure quality. It’s a safe way to refine the AI before customers interact with it. Simulations are great, but they need an evaluation step to measure success. Simulations test if the AI agent still performs as expected after changes, like updating prompts or adding use cases.

We use past real conversations to see how the AI handles them, as well as simulated scenarios, including challenging cases like dealing with angry customers, attempted tricks, or even hacking attempts. Thousands of simulations are run, and only if the evaluations are satisfactory does the AI agent go live.

Sheri Greenhaus: Is there anything else our audience should know that we haven’t covered?

Malte Kosub: One important topic is how the role of human agents is evolving. We believe that over the next year, human agents will move more into second- and third-level support as AI takes over first-level tasks.

Additionally, human agents will increasingly function as supervisors, overseeing 50 to 150 AI agents. Instead of directly taking over unresolved issues, human supervisors will guide AI agents in real time. For example, when an AI agent encounters an edge case, it consults the human supervisor, who provides specific instructions. The AI then goes back to the customer with the solution.

This shift elevates the role of human agents to AI supervisors, changing their responsibilities and skillsets.