Google builds elite team to close the coding gap with Anthropic

2026-04-20 · Source: The Decoder · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

Google Deepmind has established a specialized team, led by engineer Sebastian Borgeaud, to enhance the coding capabilities of its Gemini models, particularly for complex, long-horizon tasks like developing new software. This initiative follows an internal assessment indicating that Anthropic's programming tools currently surpass Google's offerings. To address this gap, Google is intensifying the training of its AI models on proprietary internal code, monitoring employee engagement with internal coding tools such as "Jetski," and, in certain teams, mandating AI training. Google co-founder Sergey Brin is actively involved, emphasizing that improved coding skills are crucial for developing self-improving AI and automating AI research and engineering tasks.

Key takeaway

For CTOs and VP of Engineering evaluating AI development strategies, Google's aggressive push to enhance Gemini's coding capabilities through internal code training and mandatory AI adoption signals a critical shift. Your teams should explore leveraging proprietary internal codebases for specialized AI model training to gain a competitive edge in complex software development and potentially accelerate internal R&D.

Key insights

Google Deepmind is prioritizing Gemini's coding prowess to surpass competitors and enable self-improving AI.

Principles

• Internal code training improves specialized AI models.
• Agentic execution is key for advanced AI development.

Method

Google is training Gemini models on internal code, tracking internal tool usage (e.g., "Jetski"), and mandating AI training for engineers to enhance coding capabilities.

In practice

• Train models on domain-specific internal datasets.
• Implement internal usage tracking for AI tools.
• Mandate AI training for engineering teams.

Topics

• Google DeepMind
• Gemini Models
• AI Coding Capabilities
• Anthropic Programming Tools
• Self-Improving AI

Best for: CTO, VP of Engineering/Data, AI Architect, AI Engineer, AI Scientist, Director of AI/ML

Related on AIssential

• Sergey Brin commits DeepMind to a Claude catch-up — Improving AI's coding ability is seen as the shortest path to achieving self-improving AI systems.
• Google I/O showed how the path for AI-driven science is shifting — Google is shifting its AI for science focus from specialized tools to agentic, LLM-based systems for autonomous research.
• OpenAI VS Anthropic — Frontier AI labs are rapidly advancing generative coding, focusing on agentic systems and self-improvement.
• Google's take on openclaw — The AI landscape is rapidly evolving with major model releases, strategic talent acquisitions, and significant financial shifts among key players.
• AI News: Huge Updates From Anthropic, OpenAI and Google — AI platforms are rapidly evolving towards integrated, agentic, and personalized user experiences across various devices.
• Welcome to AIE CODE - Jed Borovik, Google DeepMind — The 2025 AI Engineering Code Summit focuses on advancing the critical field of AI coding across the industry.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Decoder.