Secure AI Development
Helping a 70-person engineering team use GitHub Copilot and other AI tools with shared coding standards, automated review, and clear checks before code is merged.
Direct Wines had a 70-person engineering team shipping across multiple platforms and wanted to bring AI into its development workflow. The challenge was making AI assistance dependable across a large team: keeping generated code aligned with the architecture, applying consistent standards, and catching critical issues before they were merged. Manual code review and knowledge held by individual developers made those standards difficult to apply consistently.
We built a foundation for secure AI development within the team's existing workflow. Shared rules gave AI tools context about the codebase and engineering standards. Automated pull request review added checks before merge, with critical findings blocking approval. Codebase guidance made the same conventions available to new developers. Together, these gave the team a consistent way to use AI from the first prompt through code review.
How it works
- 01
Development workflow audit
Spent two weeks inside the team, shadowing developers and reviewing the workflow. Identified where inconsistent coding patterns, manual review, and gaps in codebase knowledge made it harder to maintain shared standards.
- 02
Shared AI rules across 4 tools
Built configuration and rule files for Cursor, VS Code, GitHub Copilot, and Claude Code. Each tool received context about the company's codebase, coding standards, and architecture, giving developers a common set of instructions for AI-assisted work.
- 03
Automated review with merge controls
Built a GitHub Actions-powered bot that reviews every pull request and ranks findings by severity. Critical issues block the merge; style suggestions remain advisory. This adds a consistent review checkpoint for AI-assisted code and helps developers focus on the findings that need attention.
- 04
Codebase guidance and onboarding
Created an AI-powered onboarding guide for questions about the codebase, architecture decisions, and team conventions. New developers can understand the standards behind the code and apply them when working with AI.
Results
- Shared coding standards and architecture context across four AI development tools
- Automated pull request checks with severity-based feedback
- Merge controls for critical findings, with advisory feedback for style
- Accessible codebase guidance for consistent AI adoption across the team
Tech stack
He quickly zeroed in on the highest-leverage opportunities and kept the team focused on what actually moves the business. He brought AI into the team in a way that actually stuck. Daniel moves fast — he goes from insight to production in days, not weeks.
Nicole Romito
VP of Ecommerce & Marketing, Direct Wines
Next step
Where does AI actually
move your numbers?Where does AI
move your numbers?
15 minutes. No sales pitch. We look at your business and tell you honestly where AI helps, and where it does not.
Book a call