# Dropbox Tech Blog null This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Testing cookie behavior across hundreds of web surfaces with our in-house auditor DevFeed: [Testing cookie behavior across hundreds of web surfaces with our in-house auditor]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good Published: 2026-08-31T17:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [Web](), [Automation](), [data]() Tags: [advertising](), [analytics](), [automation](), [browser](), [compliance](), [marketing](), [performance](), [privacy](), [testing](), [web]() ### AI overview Dropbox describes an in-house cookie auditor that acts like a privacy-conscious user, visits web pages, and checks whether loaded cookies match stated privacy preferences. The approach helps automate quality assurance across more than 200 web surfaces whose pages, integrations, experiments, and cookie behavior can change over time. ### Source excerpt Our cookie auditor acts like a privacy-conscious user by visiting web pages and checking that they only load cookies consistent with that user's preferences. ## Improving infrastructure efficiency for growing demand in the age of AI DevFeed: [Improving infrastructure efficiency for growing demand in the age of AI]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev Published: 2026-08-18T17:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [cloud-infrastructure](), [systems](), [data centers](), [Optimization](), [Hardware](), [Artificial Intelligence](), [networking](), [Magic Pocket]() Tags: [ai](), [availability](), [data-center](), [efficiency](), [energy](), [hardware](), [industry](), [infrastructure](), [networking](), [optimization](), [storage](), [systems]() ### AI overview Dropbox describes a system-level approach to improving infrastructure efficiency as AI-driven demand grows. Its engineering teams optimize capacity planning, hardware, power, cooling, storage, servers, networking, and facility design together so existing infrastructure can support growth before new data center capacity is added. ### Source excerpt As demand for AI continues to grow, so does the infrastructure needed to support it. ## How our universal content processing platform Riviera evolved for AI and beyond DevFeed: [How our universal content processing platform Riviera evolved for AI and beyond]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash Published: 2026-07-20T15:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [AI Development](), [API](), [Model Context Protocol](), [Content Management System]() Tags: [ai](), [api](), [applications](), [architecture](), [building](), [content](), [dash](), [developer](), [infrastructure](), [model-context-protocol](), [platform](), [product](), [search](), [tool](), [tools](), [workflows]() ### AI overview Dropbox describes how Riviera evolved from an internal file-preview service into a shared, large-scale content processing platform. The platform transforms content for previews, PDFs, streaming media, search, workflows, and AI applications, and its capabilities are being exposed to developers through APIs and Model Context Protocol tools. ### Source excerpt Riviera is the Dropbox content processing platform that's been iteratively improving content transformation in our products for roughly a decade. ## How we used DSPy to turn AI evaluations into better responses in Dash chat DevFeed: [How we used DSPy to turn AI evaluations into better responses in Dash chat]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon Published: 2026-06-25T16:30:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [DSPy](), [AI Chat](), [Large Language Model](), [Artificial Intelligence](), [Optimization](), [Algorithms](), [Open Source]() Tags: [agent](), [agents](), [ai](), [dash](), [dspy](), [evaluation](), [llm](), [open-source](), [optimization](), [performance](), [reasoning](), [responses](), [tools]() ### AI overview Dropbox describes using DSPy and evaluation feedback to improve Dash chat. Human-labeled examples calibrate LLM judges, whose scalable signals then optimize the chat agent's system prompt, reducing incomplete answers and token usage without compromising answer quality. ### Source excerpt We used DSPy to improve LLM judges and optimize our chat experience, creating an evaluation-driven feedback loop that produced better outputs. ## How Dropbox uses MCP and Dash to close the design-to-code security gap DevFeed: [How Dropbox uses MCP and Dash to close the design-to-code security gap]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra Published: 2026-06-12T18:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [Model Context Protocol](), [code security](), [Security](), [Code review](), [Artificial Intelligence](), [Language models]() Tags: [agentic-ai](), [agents](), [ai](), [code-quality](), [code-security](), [developer](), [developer-tools](), [engineering](), [mcp](), [review](), [security](), [workflow]() ### AI overview Dropbox built an agentic AI system combining Model Context Protocol, foundational large language models, and Dash to retrieve relevant threat models during code review. The system evaluates whether code changes align with documented security requirements and helps expose gaps between design decisions and implementation. ### Source excerpt Using an agentic AI system to surface threat models during code review and spot gaps between security requirements and implementation. ## Beyond code generation: rethinking engineering productivity in the age of AI agents DevFeed: [Beyond code generation: rethinking engineering productivity in the age of AI agents]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura Published: 2026-05-28T18:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [Artificial Intelligence](), [AI-assisted coding](), [code productivity](), [AI Development](), [Software Engineering](), [ci](), [pull-requests]() Tags: [agents](), [ai](), [ci](), [code-generation](), [code-quality](), [developer-productivity](), [developer-tools](), [developer-velocity](), [productivity](), [pull-requests](), [software-development](), [workflows]() ### AI overview Dropbox describes a shift from AI coding tools that accelerate implementation to agentic systems that execute scoped tasks. The article explains how increased code throughput creates pressure on review, CI, validation, release coordination, and production operations, requiring broader changes to engineering systems and workflows. ### Source excerpt 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. ## Introducing Nova, our internal platform for coding agents DevFeed: [Introducing Nova, our internal platform for coding agents]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura,Mike White,Kevin Altschuler Published: 2026-05-21T16:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [AI-assisted coding](), [Nova](), [AI Development](), [software-development](), [Software Engineering](), [cloud-infrastructure](), [debugging](), [test-coverage]() Tags: [agents](), [ai](), [ai-agents](), [cloud](), [coding](), [debugging](), [developer](), [developer-productivity](), [developer-tools](), [developer-velocity](), [machine-learning](), [nova](), [platform](), [software-development](), [test-coverage](), [workflows]() ### AI overview Dropbox introduces Nova, an internal cloud platform for running coding agents. The platform supports parallel coding sessions and automated workflows across the software development lifecycle, including debugging, dependency updates, test-coverage improvements, and flaky-test fixes. ### Source excerpt Nova lets engineers run multiple coding sessions in parallel and lets internal systems use AI agents as part of automated workflows. ## Improving storage efficiency in Magic Pocket, our immutable blob store DevFeed: [Improving storage efficiency in Magic Pocket, our immutable blob store]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura,Mike White,Kevin Altschuler,Facundo Agriel Published: 2026-04-02T17:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [Magic Pocket](), [cloud-infrastructure](), [Monitoring](), [Availability]() Tags: [availability](), [backend](), [cost](), [data](), [efficiency](), [infrastructure](), [magic-pocket](), [monitoring](), [operations](), [process](), [scale](), [storage]() ### AI overview Dropbox describes how it improved storage efficiency in Magic Pocket, its exabyte-scale immutable blob storage system. A change that reduced write amplification increased fragmentation and storage overhead, so the team introduced a layered, adaptive compaction pipeline along with stronger monitoring and controls. ### Source excerpt By turning compaction into a layered, adaptive pipeline and strengthening our monitoring and controls, we made Magic Pocket more resilient to workload changes. ## Reducing our monorepo size to improve developer velocity DevFeed: [Reducing our monorepo size to improve developer velocity]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura,Mike White,Kevin Altschuler,Facundo Agriel,Ishan Mishr Published: 2026-03-25T17:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [developer velocity](), [Git](), [GitHub](), [ci](), [Back end]() Tags: [ai](), [backend](), [ci](), [developer-velocity](), [developers](), [development](), [git](), [github](), [infrastructure](), [onboarding](), [performance](), [scale](), [storage](), [workflows]() ### AI overview Dropbox reduced its server monorepo from 87GB to 20GB, cutting full-clone time from more than an hour to under 15 minutes. The article explains how repository growth affected developer onboarding, CI pipelines, synchronization systems, and operational risk near GitHub's 100GB limit. ### Source excerpt Monorepos will continue to grow as products evolve, but growth doesn't have to mean friction. ## How we optimized Dash's relevance judge with DSPy DevFeed: [How we optimized Dash's relevance judge with DSPy]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura,Mike White,Kevin Altschuler,Facundo Agriel,Ishan Mishr Published: 2026-03-17T17:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [AI Chat]() Tags: [cost](), [dash](), [dspy](), [evaluation](), [framework](), [latency](), [llm](), [models](), [open-source](), [optimization](), [performance](), [production](), [prompt-engineering](), [training]() ### AI overview Dropbox describes using DSPy to optimize an LLM relevance judge for Dash, aiming to improve agreement with human labels while reducing cost and improving production reliability. ### Source excerpt 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.