Introducing AI Assist in Preset Cloud

Business intelligence (BI) software is a double-edged sword. While flexible and versatile BI tools enable users to analyze and visualize vast, complex, real-time data, users may also find themselves unsure of the best ways to manipulate the data, write SQL queries, or present information compellingly.

In the ever-evolving landscape of software development, the integration of artificial intelligence (AI) and large language models (LLMs) is reshaping productivity, enhancing user experiences, and optimizing costs across all sectors. We believe that AI can also help address the challenges faced by data teams, analysts, and users of BI tools.

These opportunities inspired us to develop Preset AI Assist, a text-to-SQL functionality allowing Preset users to use plain English to write SQL queries and expedite ad-hoc analyses. But this launch is just the starting point; we are actively exploring additional use cases to expedite insights from data.

Below, we’re excited to share our first version of AI Assist and our vision in the coming months.

Expedite SQL querying and analysis

Through Preset's integration with OpenAI's GPT-3 engine, users can promptly generate SQL queries from a simple English sentence. Traditionally, proficiency in SQL and a deep understanding of data structures were prerequisites for running ad-hoc queries and creating virtual datasets. AI Assist transforms this process, making SQL query composition faster and more accessible for individuals with varying levels of SQL skills. Whether users have basic or expert knowledge of SQL, AI Assist facilitates the retrieval of 80-90% of complete queries without the need for typing or formatting SQL.

Understand data structures and save time

AI Assist can also help users identify relevant tables and columns, as well as understand how underlying tables can be joined to answer a question. While the basics of how a database works can be quickly taught to anyone with technical inclination, grasping the intricacies of a specific schema—especially without the aid of a proper entity-relationship diagram—can become more complex if you haven't designed it yourself.

Depending on users' familiarity with the data model, they won't need to invest much time in extracting table or column schemas to determine specifics like how a particular column is labeled (e.g., is a user-identifying column called username, user_name, user, or email?), or figuring out how to retrieve information from a user table. Additionally, AI Assist can help users quickly identify the criteria for joining tables (e.g., main.user_id = user.id or main.user_id = user.user_id?).

Validate queries

Data analysts and users using AI Assist can review and modify generated SQL queries, ensuring transparency and maintaining the accuracy of data analyses. We provide this level of transparency because we value the accuracy of data analyses to drive informed, data-driven decision-making.

Supported databases

Initially, AI Assist seamlessly integrates with three popular databases: BigQuery, PostgreSQL, and Snowflake. We are actively working on expanding compatibility to include additional databases in the upcoming months. Stay tuned for more exciting developments.

Future AI applications

While AI Assist is initially designed for technical users familiar with SQL and data structure, our goal is to make self-service analytics more accessible for all user types, including non-technical users. Below are some ideas we're working on to help users find insights faster.

• Chart recommendations: Based on the data generated from SQL queries, AI Assist will recommend charts from Preset’s extensive visualization library to display the data in the most compelling ways.
• AI Assistant (chat support): If users need help with Preset, AI-powered support is available 24/7. They can ask questions in plain English, and the bot can direct them to answers from Preset’s documentation and resource library.
• Chart and dashboard creation: AI will guide users through chart and dashboard building based on their needs and inquiries.
• Global search: Even if users don't know the names of charts or dashboards, or where specific data may be located, they can find relevant content by asking questions.
• Chart insights and auto-filtering: Sometimes, it may not be obvious at first glance what insights are hidden in visualizations. AI will suggest possible interpretations of the charts and apply filters automatically to help find the insights.

Try AI Assist for Free

AI Assist is available to users on the Preset Professional and Enterprise plans. Sign up for a free trial (or log into your account if you are an existing Preset user) and check out the docs to get started.

And stay tuned - we’ll have more updates to the AI functionality in the coming months.