TechPrufer
Get started
Hex
Hex logoHex · Feature trail

Data Integrations

40 posts · Apr 2022 – Jul 2026

What this trail covers

Hex has evolved its data integration and semantic capabilities by enhancing how users can bring their data and tools into Hex for seamless analytics and data operations. This includes direct connections to data warehouses, local file uploads, metadata and semantic layer tools, data cataloging and observability platforms, code sync and versioning tools, machine learning frameworks, and orchestration/workflow tools. Hex now allows users to build governed semantic models for their data, defining tr

All Hex trails
  1. How the best data teams obsess over contextThis post details how Hex customers (Ramp, Underdog, Hover) are using Hex's data integrations, specifically with dbt and GitHub, to build and manage context for AI analytics. It highlights the technical implementation of syncing dbt metadata and domain docs via GitHub Actions, using API for endorsing models, and leveraging Context Studio with suggestions and CLI for iterative context improvement. It also mentions the use of reference repos and published apps as context for the agent.
  2. Building semantic models in HexThis post provides a practical playbook for building governed semantic models in Hex. It details the process of preparing resources by curating business questions and pairing them with Verified SQL Queries (VSQs) as ground truth. It then describes how to use Hex's semantic modeling agent with a specific prompt to scaffold the model, emphasizing the importance of detailed descriptions and testing VSQs against the model. Finally, it covers creating supporting files like usage guides and referencing them in workspace context for effective agent interaction, and outlines maintenance strategies.
  3. Are Semantic Layers the Cornerstone for AI Analytics?This post argues that semantic layers alone are insufficient for AI analytics, advocating for a broader approach to context. It introduces the concept of 'context' as a more encompassing solution than semantic layers, including database descriptions, workplace rules, semantic models, and usage signals. It highlights how Hex's platform captures and governs this context through 'Context Studio', enabling AI agents to provide trusted, accurate answers by integrating internal data with external context like strategy documents and customer interactions. The post positions Hex's approach as a partnership where user interrogation of data compounds context, leading to smarter, faster, and more trusted analytics systems.
  4. Ad Hoc Queries: Definition, Benefits & Best PracticesThis post details best practices for ad hoc queries, emphasizing the importance of governed context and semantic modeling to ensure metric consistency and knowledge sharing. It highlights how Hex's platform, through features like Context Studio and Semantic Modeling, enables safer and more scalable ad hoc analysis by embedding governance directly into the workflow, moving beyond traditional dashboard limitations and supporting self-service analytics.
  5. What Is Data Federation? How It Works, When It BreaksThis post details the technical aspects of data federation, a pattern for querying across disparate data sources (PostgreSQL, BigQuery, Snowflake) without prior data movement. It explains the middleware layer, schema translation, source-specific connectors (Trino, Athena, BigQuery federated queries, Databricks Lakehouse Federation), and the query execution lifecycle including pushdown optimization. It contrasts federation with ETL/ELT/replication, highlighting trade-offs between freshness and performance. Practical use cases like cross-database analytics, regulated environments, multi-cloud setups, and exploratory work are discussed. Potential failure points such as latency compounding, cross-source join limitations, schema drift, and security coordination challenges are also detailed.
  7. Data Orchestration: How It Works And Where It StopsThis post details the concept of data orchestration, differentiating it from ETL, data integration, and workflow automation. It explains how orchestration coordinates data collection, transformation, and activation stages using DAGs. Crucially, it highlights the 'last mile' problem where orchestration alone doesn't solve issues like untrusted dashboards, metric inconsistency, or business users bypassing governed data. The post then introduces extending governance into the analytics layer through semantic layers for metric definitions, consistent access controls, lineage beyond pipeline execution, and version control for governance changes. It concludes by explaining how combining orchestrated data with a governed analytics layer reduces manual handoffs and enables effective self-serve analytics.
  8. ETL Pipelines: Architecture, Examples, and Where They BreakThis post details the architecture, examples, and failure points of ETL pipelines, covering Extract, Transform, and Load stages. It contrasts ETL with ELT and broader data pipelines, provides five detailed examples (e-commerce, financial reporting, ML feature engineering, self-serve analytics, real-time inventory), and discusses processing modes (batch, streaming, CDC). It also touches on common ETL tools and the impact of AI on ETL pipelines.
  9. What Is A Data Lakehouse? Architecture And Trade-offsThis post details the architecture and trade-offs of data lakehouses, focusing on open table formats (Iceberg, Delta Lake, Hudi) and their role in unifying data lakes and data warehouses. It explains how lakehouses enable separate compute and storage, support batch and streaming workloads, and facilitate diverse data types for analytics and ML. The post also highlights that a lakehouse alone doesn't fix analytics workflows, emphasizing the need for a robust analytics layer with a context layer for AI grounding and governance that works for both humans and AI. It discusses the importance of semantic models and observability in the context layer and advocates for centralized policy enforcement and audit trails for governance.
  10. What is dbt (Data Build Tool)? How It Works and Where It StopsThis post details the role and boundaries of dbt (Data Build Tool) within the data stack, focusing on its capabilities in SQL-based data transformations, testing, and documentation. It explains how dbt fits between data ingestion and consumption tools, highlighting its limitations in orchestration, infrastructure management, and reverse ETL. The post also discusses the emergence of the 'analytics engineer' role, the dbt ecosystem, and the trade-offs between dbt Core and dbt Cloud. It introduces the dbt Semantic Layer as a solution for metric consistency and discusses where dbt's job ends, leading to potential issues like metric drift and bypassing governed tables.
  11. What is business logic? Definition and examples for data teamsThis post defines business logic as the rules and calculations that translate raw data into organizational decision-making metrics. It differentiates business logic from business rules, explains why scattered logic erodes trust and complicates AI adoption, and outlines a progressive approach to centralization: 1. Document and endorse datasets. 2. Move logic into version-controlled transformations (e.g., dbt models). 3. Formalize with semantic models. It highlights Hex's capabilities in semantic model sync and its Semantic Model Agent for building and editing definitions, and Context Studio for surfacing governance gaps. The post also emphasizes clear ownership of metric meaning vs. implementation.
  12. What Is Semantic AI in the World of Data?This post introduces the concept of Semantic AI in the context of data, explaining how it leverages a semantic layer to provide shared, governed definitions for business terms and metrics. It details the benefits of semantic AI, such as consistent metric definitions, fewer ad hoc requests, business-friendly data access, faster onboarding, and AI readiness. It also outlines how semantic AI works through components like the semantic layer, knowledge graphs, vector embeddings, and NLP, and highlights Hex's integrations with semantic layers from dbt, Cube, Snowflake, and Databricks, as well as its own Modeling Agent for defining measures, dimensions, and joins.
  13. Semantic Modeling: How to Build a Single Source of TruthThis post details the concept and implementation of semantic modeling as a critical component of a data integration and semantic layer strategy. It explains how semantic models define business meaning for raw data, translating between technical schemas and business language. The post outlines the building blocks (entities, dimensions, measures, metrics), the hub-and-spoke architecture, and the use of compilation engines like MetricFlow to generate SQL. It provides a practical guide on how to build a semantic model by starting with contentious metrics, building on normalized staging models, adopting incrementally, validating through CI, and closing the feedback loop with tools like Hex's Context Studio. It emphasizes that semantic models are one layer within a broader semantic context, working alongside schema metadata, endorsed tables, and workspace rules to guide AI agents and analysts.
  14. What are data pipelines? How they work & how to chooseThis post details the fundamental concepts of data pipelines, explaining their purpose, how they work (connectors, transformations, orchestration, collaboration), and different types (batch, streaming, hybrid). It also outlines criteria for choosing the right pipeline, emphasizing scalability, governance, data quality, and collaboration, and positions Hex as a platform that integrates these aspects.
  15. Introducing Databricks UC Metric Views integrationThis post details the integration of Databricks Unity Catalog Metric Views into Hex. Users can now browse, query with SQL, and build data apps using metric views defined in Databricks, inheriting their governance and security. This allows for 'on-rails exploration' with consistent, trusted definitions while retaining the flexibility for data teams to query raw tables and use mixed SQL/Python. Future integrations include no-code analysis and conversational AI leveraging metric views, and authoring metric views in Hex to push to Databricks.
  16. Introducing Semantic AuthoringHex has introduced Semantic Authoring, allowing users to create semantic models directly within the Hex platform. This includes a Modeling Workbench with AI assistance for defining measures, dimensions, and joins, as well as the ability to preview changes and sync models from Git. The platform also enhances interoperability by syncing with external semantic layers like dbt metricFlow, Cube, and Snowflake Semantic Views, and plans to allow pushing Hex-authored definitions to other systems. This evolution aims to bridge the gap between speed and trust in analytics by providing a unified platform for open exploration, deep-dive analysis, and self-serve analytics powered by governed semantic context.
  17. Data platform architecture 2025: Scalable & collaborativeThis post outlines the key layers of a modern data platform: ingestion and storage, processing and transformation (emphasizing ELT and separation of transformation from metric logic), and analytics and consumption. It discusses data warehouse, data lake, and data lakehouse approaches, as well as data mesh and data fabric architectures. It also details considerations for batch vs. streaming ingestion, storage technologies (cloud data warehouses, object storage, columnar formats), and cloud vs. hybrid deployments.
  18. Introducing new Snowflake integrations with Semantic Views and AISQLThis post details the integration of Snowflake Semantic Views into Hex, allowing users to define and ingest semantic models directly from Snowflake. It also introduces the availability of Snowflake's Cortex AISQL functions within the Hex notebook, enabling analysis of unstructured data like images and text. The post outlines the setup process for Semantic Model Sync with Snowflake and highlights how these integrations enhance AI capabilities by providing governed metrics and business logic to Hex's 'Ask Magic' feature.
  19. Introducing Semantic Model Sync in HexHex has introduced Semantic Model Sync, allowing users to connect and leverage semantic models from dbt MetricFlow and Cube. This feature syncs measures, dimensions, and joins from these external models into Hex's Explore experience, enabling self-serve analytics with governed concepts. The integration is set up via GitHub, parsing semantic code directly without third-party APIs. It supports automatic discovery of joinable datasets and error-free aggregations, and allows the use of semantically-enriched Explore cells in notebooks and chaining into SQL/Python cells.
  20. How we renovated our data warehouse without interruptionThis post details the technical implementation of a blue-green deployment strategy for renovating a data warehouse without interruption. It outlines the creation of a separate working database and schemas, the use of a `working-main` Git branch, and the setup of a dedicated dbt Cloud environment for development work. It also provides guidance on configuring CI jobs for this development branch and updating `profiles.yml` for local development.
  21. How to curate your Hex Workspace so business partners can answer their questionsThis post details how to curate data connections and semantic models within Hex to improve the 'Explore' experience for non-data users. It introduces best practices for creating dedicated data connections, filtering schemas, endorsing tables for Magic AI, adding descriptive metadata, and integrating with dbt for richer metadata. It also explains how to import semantic models from dbt MetricFlow, Snowflake Semantic Views, and Cube to abstract complexity and provide predefined metrics for self-serve analytics.
  22. You can now do no-code joins in Hex — no SQL requiredHex has launched no-code joins, allowing users to join tables without writing SQL. This feature provides an intuitive UI with guardrails, suggests join keys using AI and column profiling, warns about mismatches, and ensures accurate calculations even with fan-outs. Unlike other tools, Hex's no-code joins require no upfront modeling and can be applied in real-time during analysis. It also includes built-in safety features like smart join tests and advanced SQL generation to maintain aggregations during fan-outs.
  23. Tony Avino: How HubSpot keeps data updated and trustedThis post details HubSpot's approach to building data trust and culture within a large organization, focusing on the role of analytics engineering. It discusses the four key elements of a data-driven culture (leadership commitment, data literacy, tooling, accessibility), the structure of analytics engineering teams (product core and product outcomes), and how to measure ROI for these teams through customer surveys and time-to-insight metrics. It also emphasizes treating data deliverables as products, the concept of a data value chain with clear handoffs between data engineers, analytics engineers, and analysts, and the importance of data governance, lineage tracking (using internal tools and qualitative feedback), cost management, and applying software engineering methodologies like CI/CD and peer reviews to data asset creation. The core contribution is a deep dive into the organizational and process aspects of maintaining updated and trusted data at scale, rather than a specific technical system build.
  24. Snowpark Is Your Data LibrarianThis post details the integration of Snowpark within Hex, enabling users to perform data science and analytics at scale directly within Hex's notebook environment. It highlights how Hex's SQL cells and charting capabilities seamlessly interact with Snowpark DataFrames, leveraging Snowflake's compute for large datasets without memory constraints. The post emphasizes the lazy evaluation paradigm of Snowpark and how Hex manages the query execution and previewing, allowing for efficient analysis and visualization of billions of rows.
  25. Six Reasons Hex and Snowflake Work Great TogetherThis post details the integration between Hex and Snowflake, highlighting six key reasons for their synergy. It emphasizes secure credential management for Snowflake connections within Hex, first-class support for both SQL and Python in a unified environment, parameterized SQL queries driven by Python variables, leveraging Snowflake's performance for Hex apps, and simplified Snowpark for Python integration. The post showcases how Hex enables users to query Snowflake data securely and efficiently, enhancing the overall data analytics workflow.
  26. Transforming Data Workflows with dbt's Semantic LayerThis post details Hex's integration with dbt's updated Semantic Layer powered by MetricFlow. It highlights how Hex's integration with dbt's Metadata API streamlines metadata flow for fresh data and reliable tests, and how the Semantic Layer integration provides an easy-to-use cell for consistent and governed metrics across multiple data platforms (Snowflake, BigQuery, Redshift, Databricks). The post also mentions Hex winning dbt's Innovation Partner of the Year award.
  27. Accessing Data in Jupyter with PythonThis post details how to access data from various sources (files, databases, APIs, dataset repositories) using Python within Jupyter Notebooks. It covers installation of Jupyter, Python dependencies like Pandas and requests, and provides code examples for reading data from Excel, CSV, and JSON files, highlighting common parameters and potential issues. The post emphasizes the flexibility and ease of use of these methods for data loading and initial exploration.
  28. Top Data Modeling Tools for Streamlined Business AnalyticsThis post details various data modeling techniques (ER, relational, hierarchical, object-oriented) and subsets like Data Vault and Data Warehouse modeling. It discusses the purpose and components of each, providing visual examples and explaining their use cases. The post also touches upon the evolution of data modeling tools and best practices for structuring data for analysis.
  29. Apache Iceberg and HexThis post details the integration of Hex with Apache Iceberg, an open-source table format for massive datasets. It explains how Hex can query Iceberg tables directly (using engines like Dremio as an example) and build interactive reports on top of this data, enabling collaboration on data engineering, data science, and machine learning tasks. The post highlights Iceberg's benefits such as flexible schema management, versioning, time travel, and multi-engine access, and positions Hex as a modern, collaborative interface for working with this data.
  30. Writing Your First dbt PackageThis post details how to create and publish dbt packages, providing a comprehensive guide with working code examples and an open-source repository. It covers defining package types (internal vs. external), when to write a package, basic concepts like installation and namespacing, creating a dbt package project, defining raw data using variables, considering compatibility issues with cross-database macros, using variables to control model behavior, local testing strategies, and the process of publishing a package to GitHub and dbt hub.
  31. Q&A with Bluecore’s Data Team: Part 1This post details Bluecore's modern data stack, which includes BigQuery as their data warehouse, dbt for transformations, Looker for BI, and Hex for development, exploratory analysis, prototyping, and light data engineering/pipelines. They also use Hightouch to sync transformed data back into Salesforce and other SaaS tools, and ingest data via Fivetran from various SaaS tools. The post highlights the use of Hex for client-facing bespoke analytics solutions and prototyping, informing future UI iterations. It also touches on an alerting use case where dbt flags performance issues, and Hex provides the UI for exploration and Jira API integration for ticket creation.
  32. How to Get Your Data From Postgres to SnowflakeThis post details three methods for extracting data from a production PostgreSQL database and loading it into a Snowflake warehouse: using a connector (Fivetran), using Python with psycopg2 and Snowflake's Python connector, and using Postgres command-line utilities. It provides step-by-step instructions and considerations for each approach, focusing on the technical implementation of data movement between these two systems.
  33. Hex for Analytics Engineers: Hex Powered PR ReviewThis post details how Hex can be used to streamline analytics engineering workflows, specifically for PR reviews. It introduces the concept of using Hex as an IDE for developing data transformations, allowing for more context-rich and interactive PRs. It also demonstrates how the 'dbt audit helper' can be migrated to Hex, enabling automated validation of dbt changes and providing persistent audit traceability through app snapshots. This improves the developer experience for submitters and the quality of reviews for reviewers.
  34. Building Data Apps with just SQLThis post details how Hex enables building data applications using only SQL. It highlights features like SQL Display Tables, Query Caching, Chart Cells, and the App Builder, which allow users to transform raw SQL queries into interactive and shareable data applications without writing Python or other code. The post emphasizes the ease of use and speed to value for SQL-only analysts.