Portraits: Personalized AI Coaching with Real Experts

June 5, 2025

Google Labs has introduced a new experiment called Portraits, which allows users to interact conversationally with AI representations of trusted experts. The first Portrait features Kim Scott, the bestselling author of “Radical Candor,” and offers AI-powered coaching inspired by her communication and leadership principles.

What Changed Technically

Portraits leverages advanced natural language processing (NLP) and machine learning models to create interactive avatars that can engage in meaningful conversations. Here’s a breakdown of the technical advancements:

• Multi-turn Dialogue Management: The system uses state-of-the-art dialogue management techniques to maintain context over multiple turns, ensuring that the conversation feels natural and coherent.
• Expert Knowledge Integration: The AI is trained on a corpus of content provided by Kim Scott, including her books, articles, and interviews. This ensures that the advice given aligns closely with her principles and expertise.
• Personalization: The model can adapt to user inputs and preferences, providing tailored advice and feedback based on individual needs.

Why It Matters

For practitioners in fields like management, communication, and leadership, Portraits offers a unique opportunity to receive personalized coaching from recognized experts. Here are some key benefits:

• Accessible Expertise: Users can access high-quality guidance without the need for one-on-one sessions with the expert.
• Practical Application: The AI provides actionable advice that can be immediately applied to real-world scenarios, such as managing team conflicts or delivering difficult feedback.
• Continuous Learning: The interactive nature of the tool encourages ongoing engagement and learning, helping users develop their skills over time.

Implementation Details

Portraits is built on Google’s Gemini capabilities, which include:

• Advanced NLP Models: Utilizes transformer-based models for understanding and generating human-like text.
• Contextual Awareness: The system maintains a context stack to keep track of the conversation history, ensuring that responses are relevant and coherent.
• User Feedback Loop: Incorporates user feedback to continuously improve the model’s performance and accuracy.

Architecture Overview

The architecture of Portraits is designed to handle both the technical challenges and user experience requirements:

• Frontend: A web-based interface that allows users to interact with the AI through text or voice inputs.
• Backend:
  • Dialogue Manager: Handles the conversation flow, maintaining context and generating appropriate responses.
  • Knowledge Base: Stores expert content and user data for personalized interactions.
  • NLP Models: Processes user inputs and generates responses using advanced NLP techniques.

Benchmarks

While specific benchmarks are not provided in the initial announcement, early tests have shown:

• High Accuracy: The AI accurately captures and reflects Kim Scott’s communication style and principles.
• User Engagement: Users report high levels of satisfaction with the interactive nature of the tool and the quality of advice received.

Conclusion

Portraits represents a significant step forward in personalized AI coaching. By combining cutting-edge NLP and machine learning techniques with expert knowledge, Google Labs has created a tool that can provide valuable guidance to professionals looking to improve their communication and leadership skills.