Plaid Develops AI Assistant for In-Person Operations Utilizing Google's 'Built-in AI'

Plaid, Inc. (Chuo-ku, Tokyo; Kenta Kurahashi, Representative Director and CEO) has developed an AI assistant for in-person operations utilizing the 'Built-in AI' provided by Google in Google Chrome™.

This solution was featured as a cutting-edge use case for built-in AI technology during the session 'Build new features using built-in AI in Chrome' at 'Google I/O 2026,' held in the U.S. from May 19–20, 2026.

## Overview and Background of Adoption

Balancing privacy protection with the use of AI has been a long-standing challenge in retail and customer service settings. Because interactions with customers involve highly sensitive personal information, there has been a demand for ways to utilize AI without sending conversation data to the cloud.

The AI assistant developed by Plaid uses various Built-in AI APIs (Prompt API / Summarizer API) to complete all conversation analysis and summarization processing on-device. Since conversation data is not sent to external servers, it achieves safe AI usage that complies with privacy standards such as GDPR. Key features include:

- Real-time Staff Guidance:
It instantly analyzes the conversation content and suggests 'questions to ask next' to staff. For example, it prevents missed inquiries regarding activity levels or dietary restrictions, supporting less experienced staff in delivering high-quality service from day one.

- Human-in-the-Loop Design:
AI handles information structuring and retrieval so that staff can focus on the human strengths of empathy and building trust. The design ensures that final decisions are always made by staff, improving service quality through the collaboration between humans and AI.

- Advanced Activity Search via Cloud Integration:
Based on anonymized information processed on-device, it integrates with grounding via backend Gemini and Google Search. For instance, in the travel industry, it can instantly suggest the latest and most optimal travel plans—with images—from booking platforms like Klook or Trip.com.

We believe this combination of privacy protection via on-device processing and practical service support, coupled with the human-AI collaborative design, led to its evaluation as a cutting-edge use case.

## Future Outlook

This solution will be rolled out broadly across face-to-face operational settings that require deep inquiry and professional proposals, such as travel, staffing, automotive, real estate, finance/insurance, cosmetics, and jewelry. Furthermore, we are working on real-time multilingual support utilizing Built-in AI's Translate API, aiming to provide service support in any language environment, including inbound tourism.

Additionally, we are building systems to utilize the data obtained from in-person operations in other domains such as web/app experiences, product planning, and advertising. We position the visualization of staff contributions and linking them to appropriate evaluation as a crucial value provided by this solution.