From multi-lingual AI agents to the future of voice: Google Cloud’s Naren Kachroo at TechSparks 2026

Naren Kachroo, Head of Sales & GTM AI at Google Cloud, emphasized at TechSparks 2026 that the most valuable AI applications are those that successfully complete tasks. During the 'Voice AI in action' showcase, Kachroo demonstrated Google's latest voice models, customer use cases, and a live demo for small businesses. He argued that voice will become the primary means of interacting with AI, particularly in India, where the shift will occur in local languages and accents.

Kachroo stated that the industry has moved beyond treating generative AI as a simple question-and-answer tool. Instead, the focus is now on AI that has access to tools, data, and capabilities, enabling it to take actions on behalf of users. He clarified that they are not building AI for the sake of AI, but rather AI that improves everyday living and working conditions.

Google recently introduced two voice models: Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. Gemini 3.8 Live can understand and switch between 97 supported languages within a single conversation. The choice between the two models depends on the task at hand. Gemini 3.8 Live is suitable for simpler conversational and agentic tasks, while Extended Thinking is designed for conversations requiring deeper reasoning, context switching, and multiple steps.

Both models are voice-to-voice, handling speech as input and output in a single model. Google also offers separate models for text-to-speech, speech-to-text, and live translation. Kachroo highlighted that Gemini's latest text-to-speech model performed well in internal testing with Indian languages, accents, and tonality.

Kachroo showcased several demos, including Shopify's Sidekick assistant guiding a merchant through connecting a domain in the Shopify admin, and a voice agent booking a restaurant table for eight people. The agent had to track context across multiple turns, remember preferences, and make several tool calls, a task that requires Extended Thinking.

Another demo featured Live Avatar, which pairs Gemini's live voice with real-time video generation and lip-syncing. Google made Live Avatar available to businesses through Gemini Enterprise, marking the generation of audio and video with an invisible SynthID watermark. Citi's Citi Sky assistant, built using Google DeepMind's real-time avatar technology and the Gemini Live API, demonstrates how voice AI can assist US wealth clients in making decisions about their finances.

The lower cost of these models is due to Google running the entire stack, from its TPU chips and infrastructure to the models and applications. This approach allows Google to pass efficiency gains on to customers while offering top-quality models at a lower cost. Two customers that have already implemented voice AI into their processes include Autotrader, which uses the technology for buying and selling used cars, and Equal AI, an Indian startup whose assistant handles phone calls before the user picks up.

Kachroo emphasized that the assistant can hold conversations in multiple Indian languages, including Tamil, Telugu, Malayalam, Kannada, Marathi, and others, as 98% of India's 886 million active internet users access content in Indic languages, with one in five using voice.