This version deeply optimizes intelligent application development efficiency by expanding workflow orchestration and invocation capabilities and officially launching the Text2SQL plugin. It connects underlying metadata with upper-layer orchestration assets, further lowering the barrier to building data-driven intelligent applications.
From capability invocation to data query, application development can be more connected
As intelligent application business logic becomes increasingly complex, a single Agent or standalone tool often cannot meet complete business processing needs. How to reuse validated orchestration resources, how to make different tools work together in the same process, and how to help large models understand database structures more accurately have become key considerations in application development.
Making existing orchestration resources easier to reuse
In complex application development, if Agent and workflow resources lack convenient nested invocation methods, developers need to rebuild existing logic repeatedly. For processes that have already been debugged and validated, the inability to reuse them directly also increases subsequent development and maintenance costs.
Making different capabilities work together in the same process
MCP, Skills and various tools carry different capability responsibilities, but if they can only be invoked through separate entry points, data processing tasks are easily split into multiple stages. For business scenarios that require continuous execution of multiple steps, developers also need to handle the connections between different capabilities.
Giving natural language queries more complete data context
Natural language database queries are not just about converting user questions into SQL. Models also need to understand which tables exist in the current data source, which fields those tables contain, and how these structures are related. Without metadata such as data sources, tables and fields, large models cannot fully understand database structures, and natural language SQL generation is limited by insufficient context, especially in query scenarios involving multi-table joins.
Reducing the work of building SQL capabilities from scratch
SQL generation, correction and optimization usually involve multiple stages. If the platform lacks directly usable Text2SQL assets, users need to build related processes themselves and then combine and adjust capabilities according to actual needs.
To address these issues, qKnow Professional Edition v3.1.5 updates workflow invocation, MCP metadata query and Text2SQL capabilities, further improving the orchestration and data query chain for intelligent applications.
Whatβs new in this release?
πΉ Expanded workflow orchestration capabilities: existing resources can continue to be combined
This version expands workflow invocation capabilities and supports importing and invoking existing Agent and workflow resources. At the same time, MCP, Skills and various tools can also be orchestrated through a unified workflow entry point. This means developers can organize new business logic based on existing resources and connect different capabilities into the same workflow without building the complete process from scratch every time. By combining existing resources, workflows can carry more complex processing logic, help reduce repetitive construction work, and enable validated capabilities to be reused across different applications.
πΉ MCP metadata query: adding structural information for data queries
This update enhances MCP metadata query capabilities, supports obtaining information about data sources, tables and fields, and supports executing SQL. With these capabilities, Agents can obtain more complete database structure information when handling data query tasks. Compared with understanding query requirements only from user input, combining table structures and field information from actual data sources helps provide necessary context for subsequent SQL generation. This capability provides a foundation for natural language data queries and creates conditions for combining Text2SQL with database metadata.
πΉ Text2SQL plugin officially launched: covering multiple stages of SQL processing
This version officially launches the Text2SQL plugin asset. Built on the platformβs Bot and workflow orchestration capabilities, the plugin provides multiple capabilities around SQL processing, reducing the work required for users to build related processes from scratch.
SQL generation and rewriting: Text2SQL supports SQL generation and rewriting, which can be used to convert query requirements into SQL or adjust existing SQL according to new requirements.
SQL correction and performance optimization: In addition to generation and rewriting, the plugin also covers SQL correction and performance optimization capabilities, helping users handle issues encountered during SQL writing and adjustment.
SQL comment generation and continuation: The plugin also supports SQL comment generation and SQL continuation, further covering needs related to SQL writing and improvement.
Unlike implementing each SQL function separately, Text2SQL organizes the above capabilities into reusable workflow units. Each SQL sub-capability can be used independently and can also be combined into other Agents or applications according to application needs. This organization provides a foundation for reusing SQL capabilities in different business scenarios, allowing users to combine corresponding processes according to actual needs without fixing all capabilities in a single application.
πΉ Combining Text2SQL with MCP: improving the natural language data query chain
When Text2SQL is combined with MCP metadata query capabilities, structural information such as data sources, tables and fields can be used as context input, providing more specific data structure references for SQL generation. On this basis, the platform can better support natural language to SQL and provide capability support for scenarios such as complex multi-table queries. Users can also further build customized data query and analysis assets according to their own business needs.
From workflow resource reuse to MCP obtaining database structure information, and then to Text2SQL covering SQL processing stages, this update further connects intelligent application orchestration and data query capabilities.
Final notes
qKnow Professional Edition v3.1.5 makes targeted improvements around workflow orchestration, MCP metadata query and the Text2SQL plugin:
Workflows support invoking existing Agent and workflow resources and uniformly orchestrating multiple types of tools;
MCP can obtain metadata for data sources, tables and fields;
Text2SQL covers SQL generation, rewriting, correction, performance optimization, comment generation and continuation, and supports reusing each capability as a workflow unit.
These updates provide a more complete capability foundation for building complex intelligent applications, organizing data processing flows and carrying out natural language data queries. Specific usage can be configured according to actual business needs and application flows.
In the digital economy era, enterprises have urgent needs to activate data assets and apply intelligence. The release of qKnow Professional Edition v3.1.5 is a precise iteration based on the actual pain points enterprises face when building complex business agents. Through orchestration expansion for asset reuse, structural completion of underlying metadata, and standardized encapsulation of Text2SQL plugin assets, qKnow further broadens the boundaries of building data-driven intelligent applications.
In the future, qKnow will continue to deepen its work in enterprise-level AI infrastructure and help enterprises move steadily forward on the path of intelligent transformation with more professional and easier-to-use product capabilities.
Top comments (0)