Frequently Asked Questions.

With NFON Intelligent Assistant, NFON and botario are setting new standards in business communication: companies can connect more intelligently, communicate more efficiently, and grow more sustainably. AI-powered features in the three portfolio categories Business Telephony, Intelligent Assistant, and Customer Engagement are transforming the way businesses interact. They optimize processes, enhance personalization, and ensure maximum flexibility. Thanks to their scalability, the solutions dynamically adapt to requirements and continuously evolve with market developments.

Generative AI plays a key role in the development of new communication solutions, particularly in areas such as automated speech and text processing, personalized customer experiences, and intelligent meeting assistance. It also enables improved customer interaction with personalized and context-based responses to inquiries. Intelligent chatbots provide support around the clock. Advanced AI algorithms analyze data and deliver valuable insights for decision-making. Additionally, it creates tailored services based on customer preferences and helps generate new ideas and solutions.

NFON takes a responsible approach to the use of AI by strictly adhering to ethical guidelines as per EU AI Act, data protection regulations, and security standards. There are internal review mechanisms and regular evaluations to ensure that AI solutions are transparent, fair, and non-discriminatory.

Data protection and compliance are of the utmost priority. NFON ensures that all AI-powered solutions are GDPR-compliant and that customer data is processed securely. The embedded AI models are not trained with customer data. Details about the data processing for the chatbot Nia are available in our data privacy policy: Datenschutzerklärung | NFON

Within AI-based solutions, hallucinations cannot be completely excluded, but NFON takes measures to minimize them. This is achieved by using verified, reliable data sources, continuous quality control, and a combination of AI-driven and human validation. The AI models and algorithms are regularly benchmarked and optimized to ensure the accuracy and reliability of the generated content. In addition, users see in the respective user interfaces an indication whether AI is used (with suitable) – to ensure that users know how to work with AI related information. 

AI encompasses a wide range of technologies, and its role at NFON continues to expand. Below are some key AI-driven technologies currently in use:

·       NIA, NFON’s intelligent assistant available on the website and partner portal, is built on Botario and powered by Large Language Models (Generative AI). To ensure high accuracy in responses, it combines Prompt Engineering, Retrieval-Augmented Generation (RAG), and intent-based AI capabilities leveraging NFON’s knowledge base.

·       Botario also features an optimized Speech-to-Text (STT) model, which enables voicemail transcription in NFON’s business telephony. This model includes language detection for enhanced accuracy.

NFON continues to refine and integrate AI technologies to optimize communication and customer experience.

Botario’s AI functions can be deployed either in the cloud (SaaS in German data centers) or self-hosted on the customer’s own IT infrastructure. Self-hosting offers maximum control and data sovereignty, ensuring full compliance with internal security policies. The choice between SaaS and self-hosted deployment depends on an organization’s requirements regarding data protection, IT resources, and operational costs.

For cloud deployments, external LLM providers such as OpenAI or Microsoft Azure can be used, adhering to high security and compliance standards, including GDPR.

For AI capabilities in NFON business telephony and Nia for customer and partner inquiries – these are exclusively available with cloud deployments, with data privacy and security in focus.

Bias in AI is an important concern, and NFON is committed to responsible AI practices. However, for the current use cases bias is not a significant issue, as these systems do not process or make decisions based on personal data. Nia, which serves as knowledge bot, relies on predefined company knowledge bases, ensuring factual and neutral responses, while voicemail transcription simply converts speech to text without interpretation or decision-making that could introduce bias. While NFON continuously monitors AI developments, the current applications are not inherently prone to bias-related risks.

Yes, NFON has established ethical guidelines for AI use, outlined in the Code of Ethics and AI Policy. These guidelines emphasize integrity and compliance, transparency, risk management, and employee training to ensure responsible AI use. Any new AI applications require consultation with the AI Competence Center and Data Protection team to ensure legal compliance.

The role of call center agents is significantly evolving with the use of AI, but they will not become obsolete. Here are some key aspects of how their role is changing: 

·       Automation: AI handles repetitive inquiries, allowing agents to focus on more complex tasks. 

·       Increased Efficiency: AI-powered tools provide real-time information and reduce processing times. 

·       Greater Customer Focus: Agents benefit from AI support and can concentrate more on personal interactions. 

·       Complex Issues: Human agents remain essential for challenging cases requiring empathy. 

·       Future Outlook: Automation may reduce agent numbers, but human expertise remains crucial for complex customer concerns.

The voicemails would remain until limit has been used up (100 VM limit), once they reach the limit, the oldest one will be removed and the newest one will be entered.

Queues and Groups do not allow call waiting feature to be active. Call waiting will only work when you receive a call direct on your DDI or extension number.

Depending on the configuration, the call can be held in a queue (set with appropriate MOH and recordings) or you can set the call to go to different destination after a time out delay.

The Group and Queue configuration takes precedence over any extension members’ configuration.