At a high-energy hackathon devoted to groundbreaking AI solutions, we thrived in the fast-paced mix of collaboration and creative pressure. In just two days, we brought to life an AI voice sales agent that does what most teams simply do not have time for: calling potential customers, delivering a compelling pitch, answering questions, and, when interest sparks, guiding the conversation straight toward the next step: a proposal or a meeting. All of this, completely on its own.
Guess what we named our agent? No, it’s not Bond, and it’s not Bourne. Try again 😉
The Challenge That Sparked Our Idea
At CorvusPay we live and breathe payment solutions – and we know that many merchants, especially small and mid-sized ones, still do not offer their customers the option of paying online by card. At the same time, sales teams can not reach every potential client, and without automation, phone-based sales is becoming increasingly costly and less effective.
The result? Untapped data, underused market potential, missed opportunities and slow sales cycles. That is why we set out to create a solution that tackles all of these challenges. It is not just another tool that sits idle waiting for someone to use it, but an AI that takes full initiative from the very first moment.
Meet Edgar – The AI Voice Sales Agent
Edgar is a smart, voice-powered AI sales agent that:
- Automatically reaches out to merchants,
- Delivers a personalized pitch (for example, an online card payment service),
- Answers questions in real time and
- Steers the conversation toward the next step, whether that is a proposal, a meeting, or another follow-up.
Unlike traditional bots, Edgar does not rely on rigid scripts. His strength lies in conversational intelligence, learning from data and adapting to every single customer.
How Does Edgar Actually Work?
While it might sound simple, Edgar’s “brain” is a fusion of sophisticated technologies we integrated in just 48 hours:
- Edgar uses RAG architecture, LangChain, and FAISS to prepare its knowledge base and retrieve precise information about our services.
- Edgar pulls merchant contact data directly from our CRM.
- Edgar makes real calls to real people via the Twilio platform.
- Edgar holds natural, context-aware conversations powered by OpenAI’s GPT-4o.
- Edgar converts speech to text with OpenAI Whisper.
- Edgar speaks back in a natural human voice via ElevenLabs, with customizable gender, language and tone.
As a result, at the end of a call, Edgar can schedule a meeting, send a proposal and log the conversation into the CRM. Every call is stored as a transcript and analytics report, creating a valuable foundation for high-quality future interactions with each merchant.
Edgar in Action
During the hackathon, we prepared several demo scenarios:
- Successful outcome: Edgar smoothly steers the conversation toward a clear goal – a proposal or a meeting.
- Uninterested prospect: He detects the lack of interest and ends the call professionally.
- Off-topic detour: Edgar recognizes when the discussion drifts away from the main subject and offers a follow-up with the right person.
In every case, Edgar handled the calls entirely on his own, without a single team member having to pick up the phone.
Key Benefits of Edgar
- Personalized approach – Edgar speaks in context, never from a script.
- 24/7 availability – works when people can not.
- Scalability – can reach hundreds or even thousands of merchants in parallel.
- Interest detection – understands who is ready for the next step.
- Analytics & learning – every call makes him smarter.
- Adaptive communication – adjusts to language, region, and tone.
- Seamless integration – connects effortlessly with internal systems like CRM.
What We Learned – and What is Next
The hackathon showed us just how quickly ideas can turn into real solutions when a clear business need, a unified team and powerful technology come together.
In this first version, Edgar was a bit slow, taking two to three seconds to respond, so speeding him up is a priority. We also plan to test as many languages as possible, starting with those spoken in the countries where we operate. Integrations with additional applications, such as billing, WhatsApp, Viber, SMS and email are also on our roadmap.
For us, this AI twin is more than an experiment – we believe it has genuine commercial potential for sales teams looking to scale without increasing costs. This journey does not end with a pitch; both Edgar and our team will continue to grow and learn.
We are grateful to the organizers, mentors and fellow teams who inspired us along the way.
Edgar