Pichai: Google Lags Rivals in Agentic AI Coding

Google CEO Sundar Pichai states the company is currently behind rivals in some areas of AI-powered coding. He emphasized that Google is actively working to close this gap. Pichai noted that Google is not leading in advanced coding tasks like agentic coding, instruction following, and tool use. He explained that while Google is strong in generating one-time web interfaces, competitors have an advantage in more extensive coding tasks involving large codebases. Pichai attributed part of this challenge to a lack of developer-facing products for continuous coding interactions. He cited the relationship between Anthropic and Cursor as an example of rivals having stronger data loops. Google is addressing this with Antigravity 2.0, a new desktop application for agent-based coding workflows. Internal usage of these systems at Google is rapidly increasing. Pichai also addressed early criticism regarding the pricing, quality, and usage limits of Gemini 3.5 Flash, which is now Google's default model for AI Mode globally. He acknowledged that some restrictions were initially introduced to prevent service disruptions and that some model performance issues could be quickly fixed. He stated that AI coding is a rapidly evolving space, and Gemini 3.5 Pro is expected to roll out next month. This matters because it indicates the competitive landscape and ongoing development in AI coding tools.