Goals
Q2 2026 objectives
Objective 0 – Make the MCP a first-class, competitive data interface (Marius)
Motivation: The MCP should be a complete alternative to the UI, enabling users and agents to explore data, run analyses, and produce outputs end-to-end.
What we’ll ship:
- End-to-end querying workflows: Create, run, iterate on, and debug SQL/HogQL queries with strong error messages, metadata, and performance guidance.
- Data discovery & context: Explore connections, schemas, properties, and other metadata. Provide guidance on which connections/tables/etc to use.
- Cross-database querying: Seamless querying with HogQL across ClickHouse, Postgres, DuckDB, and other sources with consistent ergonomics
- Notebook & workflow integration: Full notebook support via MCP, including creation, editing, and async/background workflows
- Analysis & intelligence: Using notebooks, answer business questions, run statistical analyses, detect inconsistencies, and optimize queries beyond syntax
- Visualization & reporting: Generate charts, choose appropriate visualizations, and create reusable outputs like dashboards and reports
Objective 1 - Build a BI experience that feels competitive and joyful (Marius)
Motivation: A competitive BI product and a joyful SQL editor are useful products on their own, in addition to being the main building blocks of notebooks.
What we’ll ship:
- New chart types, prioritizing the most-requested visualizations
- SQL editor reliability and UX improvements
- Better variables, including multiselect, query-backed and time-based variables (e.g. for dashboards)
- Easier schema exploration and an alternate “no-sql” query building interface (manual controls + AI integration).
- Ability to click in and out of rows to explore deeper.
Objective 2 - Make notebooks the default way to analyze data (Anna)
Motivation: Notebooks, especially with Python, provide a more complete BI experience than a standalone SQL editor. We are 80% done here, just need to finish the last 20%.
What we’ll ship:
- Python notebooks for customers
- Core notebook UX and stability improvements
- Better integration between notebooks, HogQL, DuckDB, and external databases
- Initial collaboration improvements for notebooks (lock nodes if someone is editing, etc)
- A clearer default analysis flow centered on notebooks
Objective 3 - Build a first-class querying and DuckDB experience (George)
Motivation: We want a great experience for our first-party DuckDB warehouse. Today, querying and connections are inconsistent and confusing.
What we’ll ship:
- First-class DuckDB support (not treated as “just Postgres”)
- Better query errors so failures are understandable and actionable
- Improvements to the HogQL type system to support earlier errors
- Calculated properties and lambdas on top of HogQL tables
- Improved UX for connecting to databases