Telviva launches intelligent digital agent ‘Viva’ to fast-track enterprise AI adoption for SA businesses

Telviva, a leader in cloud-based unified communications, has announced the launch of Viva, an intelligent voice and text-based digital agent solution designed to make enterprise-grade AI practical, affordable, and accessible for South African businesses. Built and hosted locally, Viva marks the next chapter in Telviva’s journey of enabling quality business conversations by transforming how companies route interactions and unlock organisational knowledge.

According to Telviva CEO David Meintjes, the launch addresses a critical market need: moving past traditional, rigid chatbots to deliver integrated, autonomous customer experiences without losing the crucial human touch.

“A chatbot operates mostly in a rigid, predefined structure that mostly results in a poor customer experience; a digital agent is dynamic and easier to manage and makes the transition to a human when needed easier,” says Meintjes. “It was vital for us to develop an agent that understands natural language while connecting deeply to real business knowledge sources. This enables Viva to resolve customer queries end-to-end, without human intervention, while ensuring a seamless handoff to a human agent the moment a query requires it.”

Elevating the receptionist to ‘Customer Experience Influencer’
A core philosophy behind Viva is the evolution of traditional workplace roles. Meintjes explains that the conventional role of the front-line receptionist is fundamentally shifting into that of a Customer Experience (CX) Influencer.

By allowing the digital agent to handle repetitive routing and baseline queries, the receptionist steps into the role of “human in the loop” and focuses on the customer experience, which is a higher-value, strategic role. 

Not only do they handle complex human interactions that require deep empathy and judgement, but they also act as the quality controller of the digital agent itself. They review and assess the context of engagements, optimising the customer experience and ensuring the AI continuously aligns with the company’s service standards.”

“Our philosophy is digital when you want it, human when you need it. We are layering intelligence onto a reliable infrastructure that we have spent two decades building, giving businesses a practical and scalable approach to enterprise-grade AI.”

A scalable journey: The four agent tiers
Recognising that no two businesses are at the same point in their digital adoption, Telviva has structured its Digital Agent solution into four scalable tiers. This allows companies to start with core routing or static knowledge bases and seamlessly grow into full system integration:

• Viva Starter – Digital Receptionist & CX Influencer (Voice Only): The Digital Agent answers inbound calls 24/7 in natural language, checks user presence, and routes callers to specific people or departments. This instantly alleviates high inbound volumes, allowing human receptionists to transition into CX Influencers who oversee system quality and complex escalations.
  
• Viva Business – Text and Knowledge Agent: The Digital Agent handles text-based traffic across WhatsApp and web chat, drawing accurate responses to queries directly from the company’s website and/or an uploaded FAQ document.
  
• Viva Professional – CRM Knowledge Management Module (KMS) Agent: The Digital Agent plugs directly into a business’s existing CRM knowledge base as its primary source of truth, ensuring both the digital agent and human teams deliver identical, consistent answers.
  
• Viva Enterprise – Live Integration Agent: The most advanced tier, enabling the Digital Agent to perform real-time queries in CRM or ERP systems to deliver live, personalised details such as account balances, order statuses, or contract dates.
  

Measuring what actually matters
Telviva evaluates digital agent performance by focusing on task success, context, and user experience through four core metrics:

• Intent Accuracy Rate: Tracks if the agent correctly understood the user’s intent and responded appropriately.
  
• Task Success Rate: Confirms that the interaction achieved a successful resolution via backend execution.
  
• Semantic Error Rate: Uses an independent “LLM-as-a-Judge” evaluator to assess response quality against user intent.
  
• Turn-Taking Latency: Measures response speed down to the millisecond to ensure smooth voice interactions.
  

Together, these metrics provide a comprehensive, real-world view of agent effectiveness.

Purpose-built for the South African reality
While global AI tools are often built for global markets, Viva is tailored specifically for the local operational landscape. It natively manages Voice and WhatsApp for Business—South Africa’s de facto business communication channels—and is offered at a predictable, fixed monthly fee for unlimited simultaneous interactions, removing the budget volatility often associated with AI consumption models.

//Staff writer