Nexona Labs

BACKGROUND

Please describe your company and position.

I am the Founder & CEO of Valyrian Voice

Describe what your company does in a single sentence.

Valyrian Voice is an AI voice automation platform that provides businesses with AI-powered voice agents for inbound and outbound calling. We primarily serve businesses such as BPOs, call centers, sales organizations, and other companies that need to automate large volumes of phone conversations. Our platform is designed around a cost-efficient, low-latency voice AI infrastructure, allowing businesses to run large calling campaigns while reducing the operational cost and manual effort associated with traditional calling.

OPPORTUNITY / CHALLENGE

What specific goals or objectives did you hire Nexona Labs to accomplish?

• Develop a custom AI voice-agent system that could operate at significantly lower cost than existing per-minute voice AI platforms.
• Build an integrated STT, LLM, and TTS pipeline capable of delivering low-latency voice conversations.
• Integrate the AI voice system with real-world telephony infrastructure, including SIP and calling workflows.
• Build an orchestration system capable of managing large-scale outbound calling campaigns and concurrent calls.
• Create campaign, lead list, lead, retry, callback, call-status, and disposition management functionality around the AI voice system.
• Build a production-ready platform that could initially support our own call center operations and eventually be expanded into a multi-tenant platform for other businesses.

SOLUTION

How did you find Nexona Labs?

Referral

Why did you select Nexona Labs over others?

• Good value for cost
• Proactive problem-solving, thorough technical research, and a solution-first approach.

How many teammates from Nexona Labs were assigned to this project?

2-5 Employees

Describe the scope of work in detail. Please include a summary of key deliverables.

Nexona Labs was responsible for designing and developing the complete technology behind our AI voice calling platform. The project began with research and architecture rather than immediately moving into development. Nexona first studied the existing voice-agent market and the architecture used by major providers. They identified that the conventional approach of chaining separate third-party STT, LLM, and TTS APIs created significant variable costs, additional network latency, and increased dependence on external providers. 
Based on this, we worked together on an architecture using an integrated AI voice pipeline running on dedicated GPU infrastructure. The first stage focused on building the AI voice agent itself. Within approximately one month, Nexona delivered a working prototype that allowed me to have an end-to-end voice conversation with the AI through a web application. The next stage focused on building the telephony and orchestration layer around the AI. 
This included campaign management, lead lists, lead management, SIP integration, AI configuration, call status management, dispositions, retries, callbacks, and concurrency management. One of the more complex components was the dial-session system. It allows campaigns to be configured with selected leads, concurrency limits, AI configuration, SIP/trunk configuration, retry policies, live state, and active calls. The orchestration engine uses a sliding-window approach to keep available call capacity utilized rather than waiting for slow batches to complete. Within approximately three months, we had the core MVP, and within approximately six months, we had a fully working platform suitable for our operational use. After successfully using the platform internally, we expanded the project into a multi-tenant platform so we could offer the technology to other businesses through Valyrian Voice. 
Nexona subsequently continued supporting the platform and also addressed security improvements when I raised concerns around how tenant lead data was being stored, implementing end-to-end encryption between the tenant and server.

RESULTS & FEEDBACK

What were the measurable outcomes from the project that demonstrate progress or success?

The most significant outcome was that we were able to reduce the operating cost of approximately 1,000 outbound calls from around $200–300 per day using VAPI to approximately $10.70–11.50 per day using our own platform. At the new cost level, approximately $10 of the daily expense was telephony/SIP and only around $2 was associated with the AI infrastructure. We also achieved approximately 600–800 ms response latency, compared with roughly 3 seconds in the multi-provider architecture we had previously experienced. 
From an operational perspective, the platform allowed us to make substantially more calls than our team could manually handle, automatically detect voicemails, manage retries and callbacks, and significantly reduce manual effort. Most importantly, our successful lead outcomes improved from approximately 5 successful leads per 1,000 calls to around 15–20 successful leads per 1,000 calls. The project also progressed beyond our original internal use case. After seeing the capabilities of the platform, we decided to turn it into a multi-tenant commercial service, which is now being offered to other businesses through Valyrian Voice.

Describe their project management. Did they deliver items on time? How did they respond to your needs?

The project management was one of the strongest parts of my experience. I did not have to repeatedly ask for progress updates. The Nexona team proactively communicated the status of development and reached out whenever they needed my input, feedback, or a decision on an important product direction. They were also very cooperative when requirements changed. What I particularly appreciated was that they did not blindly implement every instruction I gave them. When they believed there was a better technical approach, they explained the relevant context, discussed the trade-offs, and gave their own recommendation. 
I did not establish strict development deadlines because I was not experienced enough with software project estimation to set meaningful timelines. However, receiving a working prototype within approximately one month, an MVP within approximately three months, and a fully functioning platform within approximately six months was very impressive to me. Whenever we encountered technical challenges, they researched the issue and worked through different approaches with me rather than treating problems as simply my responsibility as the client.

What was your primary form of communication with Nexona Labs?

• In-Person Meeting
• Virtual Meeting
• Email or Messaging App

What did you find most impressive or unique about this company?

The thing I found most impressive was that Nexona did not approach the project as a simple “build what the client asks for” engagement. I initially came to them with a cost problem. Instead of immediately accepting the existing market architecture, they researched why the existing solutions were expensive and explored whether the underlying technology could be approached differently. They were also willing to enter a technically unfamiliar area. Telephony was not the area where they had the most prior experience, yet they took on the challenge of integrating AI with SIP, building a large-scale call orchestrator, managing concurrent calls, and creating the operational system around the AI. Their ability to combine technical execution with practical problem solving is what stood out most to me.

Are there any areas for improvement or something Nexona Labs could have done differently?

Overall, I have been very satisfied with the engagement. One area that could have been improved was having even more formal documentation and planning around certain architectural decisions as the project evolved. Because the product grew substantially from an internal calling system into a multi-tenant commercial platform, some requirements and architectural decisions naturally evolved during development. That said, the Nexona team was responsive whenever these issues came up, and I consider this a relatively minor improvement rather than a significant problem with the engagement.