May: Managed tables & charts, data extraction with AI

Summary

• 🪄 Managed tables: If you wanted an editable table on top of a data source, you had to manually create actions for both retrieving and updating the data. Now, it’s as simple as selecting the table, and everything else happens automagically!
• 📈 Managed charts: The same functionality we extended to charts! Just choose your source table and select the desired columns to plot - et voilà 🪄.
• 🤖 Extract structured data with AI: In addition to direct interaction with the OpenAI APIs, you now have the ability to extract structured data from unstructured text without any code. Just provide a few examples, and let the robot do its magic!
• 🔎 Table detail view: To view a single table record, you can now open it in a convenient side panel - no need to create anything from scratch. Works for adding and editing records, too!

‍

Let's break it down

Managed tables

In the past, connecting tables to external data sources involved multiple steps and configurations. Manually creating fetch actions and adding extensive code to enable editable tables were time-consuming tasks. Now, you can effortlessly select any existing table or sheet from MySQL, PostgreSQL, BigQuery, Snowflake, MongoDB, or Google Sheets (the latter for now read-only). Adding, updating, or deleting records in a table works seamlessly out-of-the-box, with all the necessary data source operations handled automatically. Enjoy a brief introduction here 🍿:

Managed charts

During the development of managed tables, we thought: "Why shouldn't connecting data to a chart be just as straightforward?" And indeed, it should be! That's why we have added the same mechanics for the chart component. All you need to do is select the desired source table or sheet, choose the fields you want to plot, et voilà!

Extract structured data with AI

Since the initial release of our OpenAI integration, we have been amazed by the incredible apps that have been created, utilizing the capabilities of large language models. One particular use case that caught our attention is the extraction of structured data from unstructured text. Whether it's extracting complaint reasons from customer reviews or dates from hotel booking requests, AI proves to be an immensely powerful tool to automate tasks that required human attention before. To simplify the realization of such use cases, we have introduced a dedicated mode for OpenAI actions: Text to structured data.

With this targeted operation, you can define the desired fields and their corresponding types for extraction. By providing a few examples, all the complex prompt generation is handled automatically. Let me show you how it works 🍿:

Table detail view

The Table component plays a crucial role in many apps. However, when cell values are lengthy or tables have numerous columns, it can be challenging to gain a complete overview of a record's data. To address this, we have introduced the "detail view" feature, which is now enabled by default for all tables. When hovering over a row, you'll notice a small icon at the beginning of the row. Clicking on it will open the corresponding record in a user-friendly side panel, providing a comprehensive view of the data. Additionally, if the table is editable, you can modify values directly within the detail view panel!

Additional improvements

• Table column tooltips: Need to explain the semantics of a table column to your team? Just add a small description to the new tooltip property in the column configuration. A neat info-icon will appear, showing your description on hover!
• Action editor canvas: Previously, the bottom panel in the app editor would overlap with the UI canvas, causing difficulties in viewing components located at the bottom. The canvas will now dynamically resize, granting you unhindered access to the entire UI. Celebrate the freedom 🥳!
• Preserving state of data explorer: When developing an app, it's very common to frequently switch between the UI canvas and the data source explorer in the app editor. To make this more efficient, the data source explorer now remembers your exact position when switching away, ensuring a seamless transition to where you left off when coming back.
• Bug fixes & improvements: On top of all the features mentioned above, we applied 27 additional bug fixes and improvements 🚀.

‍

We are proud of these new additions to our platform and hope that they enable you to build powerful tools for your teams with rapid speed. We encourage you to share your feedback with us, whether it be positive or downright ecstatic 😜.

Happy building!
Dom