March 2025 Summaries

At Hex, their mission is to bring everyone together with data. They launched Explore last year to provide a first-class workflow for exploring and visualizing data without code. However, they recognize that there's still some wrangling involved in finding the right table and figuring out the best visualization. To address this, they're introducing Ask Magic in Explore, a conversational AI interface deeply integrated with Hex, their data, and their workflow. With Ask Magic, users can type their question on the home page of their workspace, and it will surface relevant published data apps or create a first draft of an analysis from scratch. The interface will automatically generate a summary of the insight, highlight what data was used, and explain how the analysis was generated. Users will be able to dig deeper into existing explorations by joining new data, building charts, adding filters, and generating calculations in response to requests in natural language. Ask Magic will also enable users to ask follow-up questions directly from a published app while maintaining the context of the original analysis. The tool aims to free up the data team's time for high-impact work by providing an organization's first line of defense for quick questions, and it leverages governed metrics and business logic from synced semantic models.

Fran Orozco, a data leader at BloomGrowth, is leveraging Model Context Protocols (MCPs) to streamline his data development process. MCPs provide a standardized API for AI systems to access context and use tools, allowing developers to define how tools expose capabilities to AI assistants, request actions using those tools, and automatically write code to parse/extract necessary info from tool calls. Fran is experiencing significant time savings by utilizing MCPs with tools like Cursor or Claude, enabling him to interact with Hex projects through natural language in his development environment. He has open-sourced a package called `hex-mcp` that bridges Hex's existing API endpoints with the MCP standard, allowing AI assistants to understand and use them. By using MCPs, Fran's team is accelerating development times, simplifying onboarding challenges for new engineers, and orchestrating jobs using natural language, ultimately improving their workflow and interaction with Hex as a platform.

Hex, a popular app-building platform for modern data teams, has introduced Saved Views, a feature that allows users to bookmark their favorite combinations of inputs and filters on an app. This new feature enables users to create shortcuts to jump straight to the insights that matter most, eliminating the need to reset input parameters every time they visit a published app. With Saved Views, users can save specific views for different audiences, making it easier to reference relevant data in various scenarios. Additionally, editors can list saved views, allowing others to discover and use them, while conditional notifications can be tied to these views to receive alerts only when relevant data is updated. Users can also schedule their saved views to run on a regular cadence, keeping all key insights up-to-date and synchronized.

Measuring data team ROI is challenging due to the lack of clear, repeatable North Star metrics and the fact that much valuable work is done through partnerships with other teams. Instead, data leaders focus on stakeholder voices, process improvements, and tracking changes driven by data. While there's no one way to measure data team ROI, communicating some sense of impact informs budget and headcount decisions. Qualitative impact and stakeholder trust are crucial, but also look for dollar outcomes in paid marketing, churn analysis, business health metrics, and internal workflows that correlate with ROI. Tracking time-to-insight, internal traffic to data tools, and showing stakeholders the data they need can help demonstrate the team's influence on decisions and drive meaningful change. Prioritizing the biggest opportunities and highlighting the data team's work as a shared story that shapes how teams think about products can also lead to significant impact and value. Ultimately, the key is not forcing a perfect ROI number but demonstrating the data team's influence on decisions and driving tangible changes within the business.

The conversation at The Data Blueprint event centered around hiring and structuring a successful data team, with emphasis on technical skills being essential but not sufficient on their own. Curiosity, passion for data, and the ability to navigate ambiguity and reframe problems are key differentiators that set apart strong candidates. To justify tools and headcount investments, companies should focus on demonstrating value through proof-of-concepts, investing in data quality, and highlighting the importance of data-driven decision-making. As AI evolves, it's expected to free up the data team to do more strategic work, making now a critical time for preparation.

The public beta of Semantic Model Sync for Hex has been launched, allowing users to connect semantic models from dbt MetricFlow, Cube, and LookML (in private beta) to use pre-defined measures, dimensions, joins, and more. This feature aims to address the challenges faced by data teams in keeping up with infinite demand for insights while ensuring accurate and consistent results. Semantic Model Sync enables flexible tools without locking users into vendor-specific specs, allowing them to create governed metrics without redefining them every time. The integration is available on Team and Enterprise plans, and the LookML integration is currently in private beta. Semantic models encode instructions for use of data, making it easier for organizations to leverage data in a consistent and standardized way. Hex's approach to semantic modeling focuses on flexibility, interoperability, and openness, allowing users to model what's needed without getting boxed in.