Using an agentic AI system to surface threat models during code review and spot gaps between security requirements and implementation.

How Dropbox is moving from AI tools that assist engineers to agentic systems that can execute scoped tasks, and how we’re building platforms to support those workflows.

Nova lets engineers run multiple coding sessions in parallel and lets internal systems use AI agents as part of automated workflows.

By turning compaction into a layered, adaptive pipeline and strengthening our monitoring and controls, we made Magic Pocket more resilient to workload changes.

Monorepos will continue to grow as products evolve, but growth doesn’t have to mean friction.

We used DSPy to turn prompt engineering for our relevance judge into a measurable, automated optimization loop, improving task performance, cost, and how reliably it works in production.

How we train Dash's search ranking models with a mix of human and LLM-assisted labeling.

Making products like Dropbox Dash accessible to individuals and businesses means tackling new challenges around efficiency and resource use.

From Claude Code to Cursor, we're big adopters of AI coding tools at Dropbox. The early results have been promising, but there are still a lot of open questions about how to work with these tools most effectively and where they can have the most impact. To push this conversation forward, we hosted an executive roundtable at our San Francisco studio. Here's how it went.

Engineering VP Josh Clemm deep-dives into how we think about knowledge graphs, indexes, MCP, and prompt optimization using tools like DSPy.

The feature store is a critical part of how we rank and retrieve the right context across your work.

The Dropbox Intern Program is thoughtfully designed to cultivate growth, spark innovation, and build lasting connections.

Building effective, agentic AI isn’t just about adding more; it’s about helping the model focus on what matters most.

Dropbox welcomes Mobius Labs to advance Dash’s multimodal AI, integrating Aana’s efficient architecture to enhance photo and video understanding at Dropbox scale.

Learn how Half-Quadratic Quantization (HQQ) makes it easy to compress large AI models without sacrificing accuracy—no calibration data required.

Building Dropbox Dash taught us that in the foundation-model era, AI evaluations matter just as much as model training.

Learn how to configure and customize which scopes your app requests during the Dropbox OAuth 2 app authorization flow.

Learn about planned changes to the root certificates used on the Dropbox API servers, and how that will affect SDKs used to access the Dropbox API.

Learn more about our implementation of end-to-end encryption for teams, the threat model of our design and encryption algorithms, and our commitment to minimizing the risk of data loss with a team-centric key management approach.

Building on prior prompt injection research, we recently discovered a new training data extraction vulnerability involving OpenAI’s chat completion models.

Learn how to list the contents of all of the namespaces across all members of a team using the Dropbox API

Learn about changes and updates to the Dropbox API to support the latest features of Dropbox teams

Smart move uses machine learning to analyze a user’s existing subfolder structure and suggest folders where they might want to move their files. Here's how the feature was built.

The article discusses Dropbox's implementation of a machine learning model designed to identify date formats in file names, enhancing file organization and retrieval. Effective file naming is crucial for teamwork, and Dropbox's automated naming conventions feature allows users to set rules for file names, ensuring consistency and efficiency. Initially, Dropbox attempted a rule-based approach for date identification but faced challenges due to the variety of date formats used by different individuals. This led to the development of a machine learning model that accurately recognizes dates within file names. The model underwent several stages, including data annotation, tokenization, and classification, utilizing techniques such as Inside-Outside-Beginning (IOB) tagging to label date components. The machine learning model, based on the transformer architecture (specifically DistilRoberta), showed a significant improvement over the previous rule-based system, increasing the number of renamed files by 40%. To optimize performance, Dropbox implemented techniques like model pruning and quantization, successfully reducing latency during inference. Following its rollout in August 2022, the feature gained popularity, with over a million files renamed shortly after launch. Future enhancements may include the extraction of additional entities beyond dates, leveraging advanced models for even greater accuracy in file naming conventions.