Executive Roundtable Insights on AI and Engineering Productivity

Rather than simply adopting AI tooling, Dropbox focused on making sure it actually translated into real business outcomes. While actively leveraging tools like Claude Code and Cursor, the company distinguished where to adopt, where to scale, and where to build in-house, tailored to its organizational characteristics such as a massive multi-language monorepo.

To do this, Dropbox first aligned with executive leadership to elevate AI adoption as a company-wide priority, and streamlined contract approval processes to speed up experimentation. As a result, the company confirmed effects across code review, documentation, debugging, and testing, and built its own tool that detects failed PR builds and suggests fixes through the AI platform.

Dropbox tracks PR throughput per month, per engineer as a key metric. Developers who used AI coding tools more actively saw a larger increase in code shipped, and internal surveys showed a growing positive perception of AI's impact on productivity over time.

Friction felt by engineers has also decreased. By allowing teams to choose the tools that fit them, Dropbox lowered the barrier to adoption, and most developers already have at least one AI tool built into their workflow.

At an executive roundtable held on December 11, 2025 at a San Francisco studio, several tech leaders gathered to discuss AI and engineering productivity. The conversation centered on three axes:

• Measuring impact: How to measure the development productivity gains from AI and the resulting business impact
• Leadership alignment: How to align with company leadership on the pace and direction of AI adoption
• The human element: How to hire and develop talent with AI capabilities, and how to raise productivity even for non-developers

The recurring themes throughout the discussion were balance, the role of leadership, and formalization. There was consensus that productivity gains must be managed so they don't lead to quality degradation or higher long-term maintenance costs, that tech leaders need to establish norms for AI usage, and that AI capabilities should be formally reflected in career frameworks.

The remaining question was clear too: where does the additional capacity created by AI go? Dropbox is using that capacity to address tech debt, migrations, and reliability improvements, but the challenge for 2026 lies in directly connecting productivity gains to concrete product and business outcomes, and extending that rigor beyond engineering to operations as a whole.