July 2026 Summaries

Enterprises investing in AI, particularly in the financial sector, face significant challenges with their carrier layer, which can affect the performance and reliability of AI voice agents. Many AI deployment audits overlook the carrier layer, leading to risks such as latency, loss of customer context, and restricted routing control, which can degrade AI performance. The underlying SIP infrastructure and PSTN transport layer are often the root causes of these issues, not the AI tools themselves. Financial organizations are expected to invest heavily in AI, with the assumption that their carrier layers are robust enough to handle increased demands, yet this is frequently not the case. The document outlines four risk tiers for AI readiness in contact centers, ranging from critical to low risk, emphasizing the importance of having control over voice infrastructure, real-time routing, and visibility, to prevent disruptions and ensure successful AI integration. It advocates for a thorough carrier-layer audit to identify potential constraints and improve AI voice call flows, suggesting solutions like Bandwidth's Maestro TM for enhanced control and visibility.

Inbound fraud in contact centers has surged by 26%, with knowledge-based authentication (KBA) proving inadequate against AI-enabled fraud, including deepfake voice attacks. This has resulted in significant inefficiencies, costing contact centers up to 25,000 agent-hours a month. Outbound calls face challenges too, as 80% of calls from unidentified numbers go unanswered. Financial institutions are urged to adopt a dual approach to tackle both inbound fraud and outbound trust issues. This includes integrating advanced technologies like voice biometrics and real-time call authentication to streamline verification processes and improve trust. STIR/SHAKEN protocols and branded calling can enhance call credibility, but they require robust phone number reputation management to avoid mislabeling. Institutions are encouraged to adopt a unified platform that combines fraud scoring and number reputation management, offering a comprehensive view of call authentication and reputation. Such measures are crucial to bridge the trust gap, maintain customer relationships, and ensure voice remains a reliable communication channel for financial institutions.

AI chatbots are transforming customer service by balancing speed, cost, and quality, handling routine inquiries swiftly, and freeing human agents for more complex tasks. These chatbots utilize technologies like Natural Language Processing (NLP), Machine Learning (ML), and sentiment analysis to understand and respond to customer inquiries naturally, while voice-enabled chatbots rely on speech recognition and text-to-speech to maintain fluid, human-like interactions. The effectiveness of an AI chatbot largely depends on the underlying technology and infrastructure, especially in high-stakes channels like voice calls where low latency is crucial. AI chatbots offer numerous benefits such as 24/7 availability, reduced support costs, consistent answers, and multilingual support, while also collecting data to improve products and services. Successful deployment requires careful planning, starting with simple use cases, ensuring clean data, designing smooth human handoffs, and ongoing monitoring and iteration. Bandwidth's "Bring Your Own AI" approach allows enterprises to integrate their chosen AI models with robust voice infrastructure, enabling scalable, real-time conversations across multiple channels.