September update: Enhanced AI SQL assistance
TL;DR: No more bothering colleagues for SQL help - Fabi.ai's improved AI assistant now handles everything from complex joins to date formatting, plus you can chain queries for easier debugging. With new sharing features and better data formatting, it's never been easier to get from SQL question to polished insights.
Our first iteration of AI-assisted SQL editing was extremely well received, and in September we’ve put a lot of focus on the feedback we’ve been hearing.
As a data analyst, don’t know how to join those two tables you’re querying? Can’t remember how to update a date format in PostgreSQL? Can’t remember how to filter records by a certain dimension?
No need to DM your fellow data analysts or dig through that documentation! We’ve done the work to crowdsource that information for you. Now you can simply ask Fabi.ai to jumpstart your analysis, and you can even chain SQL queries for easier debugging. CTEs be gone!
🚀 Your AI SQL assistant
If you want to make an adjustment to a SQL query but don’t really know where to start, simply ask Fabi.ai and it will offer up a recommendation. From there you can quickly tweak the results and share them back with your business stakeholder.
🔗 Chaining SQL Queries
If a report requires a long, complex SQL query involving CTEs, jump into advanced analysis mode and use the output of one SQL query as the input to the next.
Other goodies 🎁
- Quick sharing: Quick share a report with the person requesting the data.
- Text cells: Leave notes in your analysis to share your findings.
- Data formatting: Data is as much about the insights as the engagement. We sweat the small things, and we’re making sure that your data looks beautiful.
- Export data: Share your analysis with your business stakeholder, and let them export the data to take it over the finish line in their favorite spreadsheet.
Quick LLM & AI learnings
LLM models are constantly changing (even if you’re not switching models), and the output from the same prompt can change, and sometimes even deteriorate.
We’ve invested a lot in the prompt pipeline, but we’ve also invested in the output validation to guarantee the output format. This follow-on pipeline validates both the code and output format and ensures that the code returns the expected results. This component is critical to a well-functioning, trustworthy AI assistant.
