Experienced Designer, Manager,
and Successfully Exited Founder
© 2025 David Adkin. All rights reserved
We set out to solve a critical challenge for new users — especially non-engineers — who struggled to create a database. Our key insight was that users who successfully started a database were far more likely to continue building and convert to a paid plan. If they couldn’t, they’d abandon the process. With the rise of AI, we realized it excelled at generating database schemas when given the right prompts. Seeing its potential, we designed an AI-powered feature to streamline database creation, making it effortless for users to take their first step in app building.
The idea emerged from collaboration with the VP of Product & myself as we explored how to leverage ChatGPT in a meaningful way for Adalo. Instead of integrating AI just for the sake of it, we focused on finding a specific problem where AI could truly add value, ensuring that the feature wasn’t just a novelty, but something that genuinely enhanced the user experience. As the design manager, I worked closely with one of our talented product designers to bring this feature to life.
AI officially arrived on the Adalo platform with these two powerful features. ‘Magic Start’ makes it effortless to begin your database. Simply describe your app, and AI will generate collections with properties and relationships — giving you a structured foundation in seconds. ‘Magic Add’ helps when you're modifying your app. Not sure how to structure a new feature? Just select a category, provide a few details, and let AI suggest the optimal database setup. These tools remove the guesswork, making database creation faster and more intuitive than ever.
The final feature streamlines database creation — just type your app idea, and AI generates a complete database schema for you. It automatically structures collections, assigns column types (text, image, location, number, relationships), and integrates seamlessly into Adalo’s format. Users instantly see their new collections and any updates to their existing database. Behind the scenes, we engineered precise prompt rules to minimize manual input, ensuring a frictionless start. Our research showed that once users had a starting point, they quickly grasped database structure — unlocking their ability to build with confidence.
This video showcases the Figma mockups we explored while refining the feature. Early concepts included a full-screen chat interface where users could describe their app and generate a database schema. We also tested different ways to guide users in defining their app’s key features for a more comprehensive setup. In the end, we moved away from a full chat experience. User testing revealed that people often typed broader app requests beyond database setup, and many wanted changes outside the feature’s scope. Instead, we optimized AI prompts behind the scenes to require minimal input, ensuring users could start quickly and grasp database fundamentals — removing a major friction point in the app-building process.
Magic Add takes database modification to the next level, using AI to seamlessly add new collections and update existing ones. To ensure accurate schema generation, we guided makers to think through the specific feature they were adding — allowing us to refine AI rules behind the scenes. This approach not only improved schema accuracy but also provided valuable insights into the most common features makers wanted. Leveraging this data, we introduced feature templates, enabling users to instantly add fully built-out features — complete with database structures and screens—accelerating the app-building process.
By guiding makers to choose specific features (1) , we could provide tailored prompt suggestions (2) and enhance AI-generated schemas behind the scenes. Even when users entered minimal or unclear inputs, our system intelligently inferred their intent, ensuring accurate database structures. Through a combination of chat-based input and structured UI elements, we optimized AI-powered creation — allowing users to generate well-formed database schemas effortlessly. This hybrid approach proved highly effective, making the process intuitive and reducing friction in app building.
This video highlights why creating a database from scratch is so challenging — and how this feature makes it dramatically easier for new makers. The toughest part is understanding what should be a collection, how to structure data, and how relationships work. Before AI, we introduced diagrams and UI improvements to simplify these concepts, but there was still a learning curve. Magic Add changes that by handling the complexity for you while reinforcing how databases work. A key UI decision was avoiding a pure chat interface; instead, AI-generated schemas appear directly in Adalo’s database UI, ensuring users not only get the right structure but also learn how to modify it going forward.
We were thrilled with the results of this project. After rolling it out through a feature flag and testing conversion rates, our initial insight proved correct — makers who used Magic Start and Magic Add were 3.8x more likely to convert to a paid plan. Given this impact, we integrated the feature into our welcome campaign automation, ensuring new users discovered it early. By making database creation effortless, we not only improved the user experience but also drove meaningful business growth.
We tailored the announcement video to highlight how this feature benefits new makers, recognizing that AI-driven assistance would be more impactful for beginners than for experienced users. We focused on telling a story that conveyed the importance of the feature, making it relatable to our community. Explaining a complex concept like this required more than just a simple demo, so we used a mix of real UI examples and conceptual visuals to effectively communicate its value and how it simplifies the app-building process for newcomers.