How Dash uses context engineering for smarter AI

Building effective, agentic AI isn’t just about adding more; it’s about helping the model focus on what matters most. When we first built Dash, it looked like most enterprise search systems: a traditional RAG pipeline that combined semantic and keyword search across indexed documents. It worked well for retrieving information and generating concise answers. But as teams began using Dash for more than just finding content—for example, asking it to interpret, summarize, and even act on what it found—we realized that retrieval alone wasn’t enough. The natural progression from “what is the status of the identity project” to “open the editor and write an executive summary of the projects that I own” required Dash to evolve from a search system into an agentic AI.

That shift introduced a new kind of engineering challenge: deciding what information and tools the model actually needs to see to reason and act effectively. This has been popularized as context engineering, the process of structuring, filtering, and delivering just the right context at the right time so the model can plan intelligently without getting overwhelmed. We started thinking about how these ideas applied inside Dash itself, including how the model planned, reasoned, and took action on a request. Instead of simply searching and summarizing results, it now plans what to do and carries out those steps. At the same time, adding tools into Dash’s workflow created new tradeoffs around how context is managed. Precision in what you feed the model is critical in any RAG system, and the same lesson applies to agentic systems. Supplying the model with only the most relevant context, and not just more of it, consistently leads to better results.

Below, we’ll walk through how we’ve been building better context into Dash.

Dash, an AI that understands your work, knows your context, your team, and your work, so your team can stay organized, easily find and share knowledge, and keep projects secure, all from one place. And soon, Dash is coming to Dropbox. Learn more →

Engineering context precision in Dash

As Dash gained new capabilities, we faced the challenge of ensuring that the model had access to the right information at the right time. This required a deeper understanding of the context in which the AI was operating. We began by analyzing the types of requests Dash was handling and the information it needed to fulfill those requests effectively. This involved identifying the key entities, actions, and relationships involved in each task.

One of the first steps was to improve the way Dash understands and interprets user requests. We implemented natural language processing techniques that allowed the model to parse and disambiguate queries more accurately. This enabled Dash to recognize the intent behind a request and determine the appropriate response. For example, if a user asked, “What are the next steps for the identity project?” Dash would need to understand that the user is looking for actionable items rather than just a list of documents.

Next, we focused on structuring the information that Dash accesses. We realized that providing the model with a well-organized and relevant dataset would significantly improve its performance. We implemented a system that automatically categorized and tagged documents based on their content, making it easier for Dash to retrieve the most relevant information. We also integrated Dash with various tools and APIs that provided additional context, such as project management software and collaboration platforms.

Another critical aspect of context engineering was ensuring that Dash had the right set of actions and functions to perform. We developed a modular architecture that allowed us to easily add new tools and capabilities to Dash’s workflow. This enabled us to provide the model with the necessary functions to carry out complex tasks, such as generating reports, scheduling meetings, or automating workflows.

However, as we added more tools and functions, we had to be careful not to overwhelm the model. We implemented a system that prioritized the most relevant actions based on the user’s request and the available context. This ensured that Dash focused on the most important tasks and avoided unnecessary distractions.

We also placed a strong emphasis on user feedback. We integrated a feedback loop that allowed users to rate the quality of Dash’s responses and provide suggestions for improvement. This feedback was used to continuously refine the model’s understanding of context and its ability to deliver relevant information and actions.

In addition to these technical advancements, we also focused on making Dash more user-friendly. We designed an intuitive interface that allowed users to easily interact with the AI and provide additional context when needed. This included features such as voice commands, chat interfaces, and integrations with popular productivity tools.

As Dash continues to evolve, our goal remains to create an AI that can understand and act on complex tasks with precision and efficiency. By focusing on context engineering, we aim to build an agentic AI that can learn, adapt, and improve over time, providing users with a powerful and intelligent tool to support their work.

In conclusion, the journey from a traditional search system to an agentic AI has been a transformative one for Dash. Through context engineering, we have been able to build a system that not only retrieves information but also plans, reasons, and takes action effectively. As Dash expands its capabilities and integrates with more tools and platforms, our commitment to precision and relevance will remain at the forefront of our engineering efforts. With the upcoming release of Dash on Dropbox, users will soon experience the benefits of an AI that truly understands their work and can help them achieve their goals more efficiently.