KDB-X webinar recap: Build, analyze, and innovate on the next generation of kdb+

5 Key Takeaways

1. KDB-X is the next major leap in a 30-year technical lineage

Manish opened the session with the story of how KX evolved from the original K and kdb,  to kdb+ and now KDB-X. Each era shaped by new demands in real-time analytics, scale, and now AI-driven workloads. KDB-X extends the proven kdb+ engine into an AI-native platform that unifies streaming, historical, vector search, and open-format interoperability in one environment .

It retains full backward compatibility with existing q/kdb+ code while introducing new capabilities that weren’t possible before, including modules, open format access, and native Python/SQL integration.

2. A unified programming model powered by q, now extensible for modern teams

A central theme was KDB-X’s unified programming model, built on q’s vector-based execution engine. Unlike specialized databases that require separate systems for streaming, SQL, and vector workloads, KDB-X provides one runtime for all real-time data operations .

A major part of the GA release is the new module system, which brings modern software-engineering practices to q:

• namespacing and isolation
• composability
• cleaner imports (use)
• standardized packaging

This makes it far easier to share, reuse, and maintain code across teams — and sets the foundation for module repositories, versioning, and dependency management coming in future releases.

3. Native support for open formats, including Parquet, unlocks the road to the lakehouse

Connor walked through how teams can now query Parquet datasets directly from KDB-X using virtual tables, without conversion or backfill. This enables:

• immediate access to existing lakehouse datasets
• seamless qSQL queries over Parquet partitions
• partition pruning, column projection, and MapReduce execution patterns
• easy exploration of large vendor or internal datasets before engineering integration

Connor showed real benchmark comparisons between Parquet and native kdb+ formats, demonstrating how developers can evaluate performance tradeoffs and integration paths themselves .

This capability lays the groundwork for upcoming support for Iceberg-style table formats and deeper object storage integration.

4. AI-native workflows with the KDB-X MCP Server and built-in AI libraries

Aldred introduced the KDB-X MCP Server, an open-source, extensible foundation for connecting LLMs and agentic workflows directly to KDB-X with full governance.

The demo showed how analysts can use natural language to:

• run SQL queries through AI-assisted tooling
• perform vector and hybrid similarity search using KDB-X’s AI libraries
• analyze structured + unstructured CRM data
• generate summaries, investigations, and portfolio reports automatically

Behind the scenes, the MCP Server handles connection management, tool execution, resources, and prompting — letting developers extend or tailor workflows using templates for tools, resources, and prompts .

This positions KDB-X as a core engine for real-time, agent-driven analytics and RAG systems.

5. A transparent roadmap focused on extensibility, interoperability, AI services, and developer experience

The session closed with a look at the public roadmap, now published on the KX Developer Center. Highlights include:

• Extensibility: module versioning, dependency management, and a central registry for publishing and discovering modules.
  Interoperability: expanded Parquet integration, deeper object storage capabilities, and future support for Iceberg-style open table formats.
• Dev Experience: improved debugging, profiling, linting, and modernized developer workflows.
• Data & AI Services: GPU-accelerated KDB-X, a Vector DB service, and a unified DB Services layer for real-time and historical workloads under a single interface .

The roadmap reflects ongoing collaboration with users and the broader community.

Start Building with KDB-X Today

Whether you’re exploring new datasets, building analytics workflows, or developing AI-powered applications, KDB-X is ready to support your next project. The Community Edition is completely free, including for commercial and offline use, and installation takes just minutes.

Visit the KX Developer Center to download KDB-X, access tutorials, read the documentation, and join the community.