Wrenai is an Open Source-Source Commercial Information Agent developed by Canner, designed to allow transparent and natural language interaction with structured data. It targets both technical and non -technical teams, providing the tools to question, analyze and view the data without writing SQL. All capacities and integrations are verified in relation to official documentation and the latest versions.
Key capacities
- Natural language in SQL:
Users can ask data questions in simple language (on several languages) and Wrenai translates them by specific SQL requests for production quality. This rationalizes access to data for non -technical users. - Multimodal output:
The platform generates SQL, the graphics, the summary reports, the dashboards and the spreadsheets. Textual and visual outings (for example, graphics, tables) are available for the immediate presentation of data or operational reports. - Genbi Insights:
Wrenai provides summaries generated by AI, contextual reports and visualizations, allowing rapid analysis and lending to the decision. - LLM flexibility:
Wrenai supports a range of large -language models, including:- OPENAI GPT series
- Azure Openai
- Google Gemini, Vertex Ai
- In depth
- Databricks
- AWS kettle (anthropogenic claude, cohere, etc.)
- Hail
- Olllama (to deploy local or personalized LLM)
- Other models compatible with the OPENAI API and defined by the user.
- Semantic layer and indexing:
Use a modeling language definition language (MDL) for coding scheme, measurements, joints and definitions, giving the precise LLM context and reducing hallucinations. The semantic engine guarantees queries rich in context, diagram interests and recovery based on relevance for a precise SQL. - Export and collaboration:
The results can be exported to Excel, Google Sheets or API for more in -depth analysis or team sharing. - Embeddability API:
Request and visualization capacities are accessible via API, allowing transparent integration in personalized applications and fronts.
Overview of architecture
Wrenai's architecture is modular and very extensible for robust deployment and integration:
| Component | Description |
|---|---|
| User interface | User interface on the web or CLI for natural language requests and data visualization. |
| Orchestration layer | Manages the input analysis, manages the LLM selection and coordinates the execution of the request. |
| Semantic indexing | Integrate the database diagram and metadata, providing a crucial context for the LLM. |
| LLM abstraction | Unified API to integrate several LLM suppliers, both cloud and local. |
| Query engine | Executes SQL generated on the databases supported / data warehouses. |
| Visualization | Rendered the tables, the graphics, the dashboards and the results of the exports if necessary. |
| Plugins / Extensibility | Allows personalized connectors, models, rapid logic and integrations for specific needs. |
Semantic engine details
- Integration of diagram:
Dense vector representations capture the scheme and the commercial context, fueling relevance recovery. - Invite a few strokes and injection of metadata:
Diagram samples, joints and commercial logic are injected into LLM prompts for better reasoning and better precision. - Context compression:
The engine adapts the size of the scheme representation according to the token limits, preserving the critical details for each model. - Retriever generation:
The relevant scheme and metadata are collected via vector research and added to invites to align the context. - Model-Agnostic:
The Wren engine operates through LLM via abstraction based on the protocol, ensuring a coherent context regardless of the backend.
Supported integrations
- Databases and warehouses:
Ready support for Bigquery, Postgresql, MySQL, Microsoft SQL Server, Clickhouse, Trino, Snowflake, Duckdb, Amazon Athena and Amazon Redshift, among others. - Deployment modes:
Can be executed self-heberged, in the cloud or as a managed service. - API and integration:
Easily integrates with other applications and platforms via the API.
Typical use cases
- Marketing / Sales:
Rapid generation of performance graphics, funnel analyzes or summaries based on the region from natural language prompts. - Product / Operations:
Analyze the use of products, the unsubscribe of customers or operational measures with follow -up issues and visual summaries. - Managers / analysts:
Automated and up -to -date commercial dashboards and KPI follow -up, delivered in a few minutes.
Conclusion
Wrenai is a verified and open source Genbi solution that fills the gap between the sales teams and the databases thanks to conversational analyzes, the context and supplied by the AI. It is expandable, multi-LLM compatible, secure and designed with a strong semantic backbone to guarantee an intelligence of confidence, explainable and easily integrated.
Discover the GitHub page. All the merit of this research goes to researchers in this project.
Asif Razzaq is the CEO of Marktechpost Media Inc .. as a visionary entrepreneur and engineer, AIF undertakes to exploit the potential of artificial intelligence for social good. His most recent company is the launch of an artificial intelligence media platform, Marktechpost, which stands out from its in-depth coverage of automatic learning and in-depth learning news which are both technically solid and easily understandable by a large audience. The platform has more than 2 million monthly views, illustrating its popularity with the public.
