THE national customer service or better known as the BPO industry in the Philippines reached P13 trillion in 2023, nearly 10 percent of the GDP. With many multinational companies outsourcing operations to the Philippines, the demand for scalable and efficient customer service solutions is strong.
The market has fragmented due to various communication platforms, with customers engaging through phone calls, social media, messaging apps like WhatsApp, Viber, Facebook Messenger and email. Contact centers face challenges in balancing diverse service modes while maintaining quality across platforms. They must also navigate the complexities of interacting with a new-age consumer seeking personalization and convenience across multiple devices.
Leading the charge in customer service
Contact centers face challenges and must now navigate the complexities of interacting with a new-age consumer seeking personalization and convenience across multiple devices. CONTRIBUTED PHOTO
The launch of ChatGPT in 2022 was a groundbreaking moment in technology history, akin to the advent of the World Wide Web. It dramatically transformed our relationship with artificial intelligence, deeply mainstreaming it into public consciousness. Over the past year, the focus has shifted to practical uses, with researchers, businesses and leaders integrating this tech into their daily workflows. With powerful new models like GPT-4o and Claude 3.5 Sonnet, we’re gradually moving toward closing the gap between human and machine understanding — a huge feat. Nowhere is this impact more anticipated than in the Philippines’ contact center industry.
When it comes to customer service, we’ve moved far beyond traditional chatbots; the spotlight is now on large language models (LLMs)-powered dynamic AI chat and voice agents. These AI agents not only handle basic level 1 customer service issues but are increasingly tackling more complex level 2 queries. Recent advancements in multimodal AI, which blend language, voice and vision, are paving the way for more natural and fluid conversations.
Moreover, today, we see interactions not only spanning modalities but also multiple regional languages. In the Philippines, where over 170 languages are spoken, this technology is game-changing for the contact center and customer service industries, supporting businesses both locally and globally.
Personalization, user-friendliness
A key element in resolving these challenges is the use of multimodal AI. Unlike unimodal LLMs, multimodal LLMs integrate various modalities, enabling models to understand inputs across different formats effectively. This capability enhances their decision-making and facilitates outputs that seamlessly integrate multiple modalities, leading to natural conversations.
Instead of merely clicking buttons and typing, we can now engage with AI at scale, sharing images and, potentially in the future, gestures. For example, when assembling furniture, explaining an issue through text can be frustrating. With multimodal AI, customers can take photos or videos of their progress and send them for support. A multimodal LLM-powered AI agent can respond with tailored assistance, such as diagrams, 3D videos, or step-by-step audio guides, maintaining the patience and empathy of a human agent.
The flexibility, precision and scalability of multimodal AI optimize customer service operations, enhancing the internal functioning of contact centers and freeing human agents for specialized tasks. This efficient resource allocation boosts employee satisfaction and reduces turnover, a common issue in support centers. Moreover, advanced models like GPT-4o enable AI agents to approach human-like interactions. Multimodal LLM-powered voice AI agents can adjust their tone in real-time, responding to user emotions such as frustration or happiness. In the contact center realm, an AI-powered voice that understands and adapts to emotions is transformative.
Bridging the language gap
The second aspect of solving the problem of personalization is bridging the language gap — something that can go miles in making the customer feel at ease. How can one make customer service multilingually accurate? Despite AI’s universality, it is clear that, so far, LLMs have a heavy bias toward English-speaking users. That’s not entirely surprising: 40 percent of models today are produced by US-based companies, with many existing models trained in the English language.
In a market like the Philippines and the rest of Southeast Asia more broadly, these LLMs have limitations due to the lack of linguistic representation of languages. As mentioned, the Philippines is home to 170 local languages while SEA has over 1,200. Multilingual capabilities are critical for organizations that cater to customers across the region speaking diverse native languages.
Unlocking that potential will require LLMs trained in the region’s languages, enabling people to communicate with machines in their native tongues while ensuring machines better understand the user’s context to generate higher-quality outputs. This is where regional LLMs like Komodo-7B and SEA-LION come into play. Their development marks a crucial step toward inclusivity and effectiveness in language models, catering to the linguistic needs of diverse communities. By leveraging such models in customer service, businesses can extend their reach within the region. For customer service units, these offer the potential of multilingual self-service, language-agnostic knowledge base, colloquial understanding, and efficient human assistance.
AI is transforming the customer service and contact center industry at a breakneck pace, marked by milestones accelerating the move toward a fully autonomous future. Among these, multimodal AI and multilingual regional LLMs stand out. The future will undoubtedly see a merger of these technologies, creating sophisticated foundation models that better understand the Southeast Asian populace and deliver unparalleled AI-first personalized customer experiences.
Raghu Ravinutala is the chief executive and co-founder of Yellow.ai, a global provider of generative AI-powered customer service automation.
Be the first to comment