DataSense

Cittabase DataSense AI is a Snowflake Native App that leverages the power of Large Language Models (LLMs) to automatically generate intelligent metadata and high-level insights for any table in your Snowflake environment. Simply provide the database, schema, and table name, and DataSense AI. It will generate the following.

✅ Data Preview & EDA – Preview your data and plot Univariate and Bivariate charts.

✅ Observations on Fields – Identify meaningful information about key fields.

✅ Other Key Factors – Other relevant information regarding the data along with any anomalies.

➢ Designed for business analysts, data engineers, and governance teams, DataSense AI accelerates data understanding and insight generation—empowering organizations to make informed decisions faster within the Snowflake Data Cloud.

➢ If Cortex LLM functions are not enabled, you could consider enabling cross-region inference by referring to the link below: ([https://docs.snowflake.com/en/user-guide/snowflake-cortex/cross-region-inference](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cross-region-inference)). This enables inference requests to be processed from a different region where LLMs are available. Please note that you will be charged credits for the use of LLM as described in detail in the above link. Please review the same. Credits are considered consumed in the requesting region.

➢ This application leverages LLM models, which may exhibit biases and potential inaccuracies based on the data they were trained on. While every effort is made to provide accurate and unbiased insights, the generated descriptions and summaries should be reviewed and validated.

**Note: *DataSense AI utilizes the Mistral-7B LLM within Snowflake and will consume Snowflake credits. Please refer to the official Snowflake Credit Consumption Table (https://www.snowflake.com/legal-files/CreditConsumptionTable.pdf) for detailed credit consumption.***

搜集汇总

背景与挑战

背景概述

DataSense是Snowflake原生应用，利用Mistral-7B大模型自动生成数据表的元数据分析和可视化洞察，帮助业务团队快速理解数据。使用时需注意跨区域推理设置、信用消耗及模型可能存在的偏差问题。