Augment unlocks complex codebases with Claude on Google Cloud's Vertex AI

Augment Code uses Claude on Google Cloud's Vertex AI to transform how developers understand and work with sophisticated codebases. It gives enterprises an AI expert that understands their software systems and helps teams navigate code confidently.

With Claude on Vertex AI, Augment has:

• Reduced project timelines from 4-8 months to 2 weeks for one enterprise customer
• Accelerated developer onboarding from weeks to 1-2 days
• Processed millions of lines of code across hundreds of customer codebases
• Implemented secure protocols with SOC 2 Level 2 certification

Bringing expert-level AI to complex software

Augment builds AI-powered software development tools that help teams work with enterprise systems like networking platforms, databases, and storage infrastructure. Current AI coding tools have limitations that might explain their modest adoption among expert developers. Augment takes a different approach. "Building an AI that's a novice with your software isn't valuable," said Scott Dietzen, co-founder at Augment. "We wanted to create an AI that becomes an expert in your codebase."

Breaking down barriers to understanding complex software

Enterprise software teams must navigate codebases with millions of interdependent lines of code, where every change can have far-reaching impacts. Igor Ostrovsky, CTO and co-founder at Augment, said, "You can't write a line of code without understanding that context. We built an AI system that combines large language models with deep knowledge of an organization's code, enabling developers to write, document, debug, and maintain software more efficiently."

Why Augment chose Claude on Vertex AI

After testing multiple AI models, Augment selected Claude on Google Cloud's Vertex AI for three main reasons:

Augment found Claude's code understanding capabilities particularly valuable. Dietzen said, "Sonnet has abilities beyond other code models." "We switched our production site to Claude in a weekend based on the quality." The team regularly tests available models using contractors who work on real-world code tasks, with Claude consistently performing well.

Another important factor was security. Augment implemented a system requiring users to verify access before viewing any codebase. Combined with SOC 2 Level 2 certification and Google Cloud's infrastructure, this helps Augment protect customer data. "Everything stays inside Google Cloud," explained Guy Gur-Ari, Chief Scientist at Augment. "This makes our security story very clean and simple to explain to users."

Infrastructure reliability also influenced their decision. Ostrovsky said, "Google's core cloud services are stable—no networking or DNS issues. With Google Cloud, we can adapt quickly to changes." Their implementation uses several Google Cloud services and Vertex AI for model experimentation.

How Claude transforms software development

Augment combines Claude with five to six other AI models to enhance development through:

• Contextual Code Understanding: Developers can chat with their codebase to quickly comprehend new areas
• Impact Analysis: The system evaluates potential changes across the entire codebase, identifying risks and dependencies
• Intelligent Code Reuse: Instead of generating new code, the system suggests existing libraries
• Documentation Generation: AI assistance for creating and maintaining technical documentation
• Change Propagation: When developers modify code, the system identifies all related areas needing updates, from databases to APIs and documentation

Delivering measurable impact

Claude on Google Cloud has significantly improved development team workflows at Augment. Gur-Ari shares, "Tasks that would take weeks for a developer to learn can now be completed in a day or two." This efficiency is demonstrated by real results—one enterprise customer finished a project in two weeks that their CTO had initially estimated would take 4-8 months.

The technology is also valuable during critical incidents. Dietzen explained, "During a 2:00 AM outage, you can ask our chat about the issue and immediately get oriented toward the solution. You can trust the answer because the AI understands how all the components interconnect."

Building the future of AI-augmented development

Augment envisions a future where AI transforms software development by elevating—not replacing—human developers. Dietzen said, "Our vision from the outset was to augment, not replace human software engineers. Humans have a deep understanding of how software should evolve to meet business needs. We want to accelerate development by removing the technical barriers that slow teams down."

Augment is a leader for this transformation in software development through their partnership with Anthropic and Google Cloud. "We're pushing the state of the art on AI capabilities," said Ostrovsky. "We're building agents that can accomplish larger tasks independently while gaining deeper understanding of organizational knowledge and data." By making their platform available to all developers while maintaining enterprise-grade security, Augment aims to give every software team the power to work at the speed and scale previously possible only at the world's largest tech companies.