Announcing MCP Toolbox Java SDK v1.0: Agentic data access for the enterprise
cloud.google.com·6d ago
Cloud & Infrastructure
The New Stack·Aug 7·5 min read
Coinbase, Shopify and Ramp all built their own coding agents. All three still pay Anthropic.
TL;DR
Coinbase, Shopify, and Ramp have developed their own internal coding agents for software development, yet they still rely on commercial AI models from Anthropic. This highlights a shift towards owning the execution environment while leveraging existing AI models for reasoning.
✦ Why It Matters
Consider developing an internal coding agent to enhance your team's software development process while leveraging existing AI models.
Key Takeaways
- Consider building an internal orchestration layer for AI tools.
- Integrate existing tools like Slack and GitHub for automation.
- Evaluate the cost implications of using AI models in your workflows.
- Leverage commercial coding assistants for interactive development sessions.
Full Summary
Coinbase, Shopify, and Ramp have each created internal coding agents to assist their developers, yet they continue to utilize commercial AI models like Claude Code from Anthropic. These companies focus on building the execution environment, or 'agent harness,' which manages context, permissions, and workflow orchestration, rather than the underlying large language models.
For instance, Coinbase's Forge automates bug fixes by integrating with tools like Slack and GitHub, while Shopify's River operates on a robust engineering platform that supports a significant portion of merged pull requests. Ramp's Inspect further enhances this architecture by executing in sandboxed environments.
This trend indicates that enterprises are prioritizing the orchestration layer to optimize costs and improve governance, treating AI agents as integral to their internal developer platforms.