Google Cloud’s no-code development service, AppSheet, offers to add functionality using spreadsheet formulas and expressions. It can also train and apply machine learning predictive models and OCR models.
Google Cloud AppSheet is a no-code application development platform. Its web-based design environment runs in Google Cloud, and it generates apps for the web (mobile or desktop browser), iOS (11+), and Android (5+). Although advertised as no-code, AppSheet supports spreadsheet formulas, filter expressions, and robots defined using expressions and charts, giving it much of the functionality of spreadsheet creators. low-code apps.
AppSheet aims to automate business processes such as order approvals and user notifications, and automatically build actions and views based on user intent with Google Cloud AI and ML. User can use AppSheet to build a single application for use on desktop, mobile and tablet. It also offers the ability to connect to a variety of data sources, as well as add data such as GPS locations, images, drawings, barcode scanning and character recognition from devices. end users.
Make IT and business work together
Google does not claim that “citizen developers” can build apps in a vacuum. Instead, the firm says IT developers and citizens can collaborate effectively through corporate governance and policy capabilities. They also tout the integration between AppSheet and Google Workspace tools as a time saver. As with most no-code and low-code app builders, citizen developers use AppSheet because it’s easy, and professional developers use it because they can build apps quickly. .
Much of the design work for AppSheet is done at the data design stage. Of course, it helps to know what the user is trying to accomplish and why. AppSheet’s process for inferring intent from data structure is quite good. The app supports many data and service integrations, including some non-traditional data sources such as a Google Drive folder treated as a table. Power Apps, which is more mature, boasts of having a lot more integrations than AppSheet. In addition to Microsoft Power Apps, Google Cloud AppSheet competes with Amazon Honeycode and around 400 other low-code or no-code app makers.
Creating an AppSheet app
Google breaks down the process of building an AppSheet app into eight steps:
– Prepare your data;
– Connect its data to AppSheet;
– Become familiar with the AppSheet editor;
– Define how its connected data will be used;
– Create views and customize the look and feel;
– Create custom buttons, actions and automations;
– Test, share and deploy its application with users;
– Improve its application and get feedback.
Essentially, Stages 1 and 2 embody the intent-driven prototype generation phase. In some cases, the prototype will be good enough to use. More often than not, the user wants to refine the application to do more and integrate more data. Over the life of the application, as it grows within the organization, data from Google Sheets may need to be migrated to a more scalable database, such as Google Cloud SQL .
Step 3 takes more time and effort than it looks. AppSheet has a surprisingly large “surface” to support extended functionality. It’s worth getting familiar with the app sheet editor so you don’t feel lost afterwards. Besides, there are more than 50 app templates to start working on common use cases.
The user can start an AppSheet application with data, with an idea, or with a model. (Credit: IDG)
Above, the usual way to start with data is to connect to a spreadsheet through a cloud storage provider. It is also possible to connect to an SQL database table or a few non-SQL data sources. (Credit: IDG)
Once the application is generated, it is possible to preview it on a smartphone, tablet or full-screen emulator, as well as modify it if necessary. It is also possible to view it from its devices. The AppSheet editor suggests additions and also allows you to make your own additions. (Credit: IDG)
The user can build an application from data, and in addition, start with a model. The AppSheet currently has about 50 starter templates. (Credit: IDG)
AppSheet can take advantage of many sensors built into smartphones and tablets. It can also integrate with many applications and platforms. (Credit: IDG)
AppSheet currently connects to 17 cloud databases and tables, and integrates with 11 external services. However, these numbers are no match for AppSheet. For example, the platform works with Zapier, which integrates with over 3,000 other apps using a no-code GUI, and with Apigee API, which allows you to create API proxies, flows, and policies. for back-end services. (Credit: IDG)
Running AppSheet apps
In addition to being able to preview your applications in the web emulator, visualization is available on supported devices. Just go to the Users pane and enter the email of the user you want to preview.
On the device, use the link in the email to download the AppSheet shell application, then follow the instructions to download the created prototype application. In order not to take a considerable time to load, it is possible to choose between data synchronization at startup or deferred (manual) synchronization.
Screenshot of the National Parks app map view taken on an Android phone. (Credit: IDG)
Define how the data will be used
As stated earlier, much of the design work for AppSheet happens when building and choosing the data sources themselves. However, refining how the application uses the data received from the source is very easy. One way to do this is to change the column specifications under Data | Columns, for example to add virtual columns with formulas, such as a tax column that multiplies the state sales tax rate by the price.
Another method is to create slices of the data source, which are essentially filtered subsets of the table. The filter is done both by rows and by columns, for example by placing each category value in its own slice and including only the most important columns.
While AppSheet infers the columns and column names from the source table (here an Excel spreadsheet), it is easy to control if the columns are displayed, editable, required, etc. Editing of formulas for virtual columns is also available, and AppSheet helps to create the formulas.
Add data sources, views, and screens
Once the application is working with a data source and as many views and screens as needed, the organization and/or its users may want to extend functionality, often by adding more data and more views and screens. screens to go with. For example, after setting up an inventory application, sellers could add an ordering function to this same application.
Create actions, bots, predictive models and OCR models
In AppSheet Editor, three types of actions can be created: UI Navigations (show a new view of this or another app), Data Changes (perform CRUD operations), and External Communications (send a push notification or text message). Additionally, system actions are automatically created. These actions must then be configured to run in response to navigation events.
Bots are a more general type of action. In general, they follow the following pattern: “if an event occurs, execute a process consisting of one or more tasks”. Essentially, bots achieve process automation. They can run in the background, and activate on a data change or on a schedule.
Since AppSheet is a Google Cloud service, it’s no surprise that it can train and run machine learning models. Currently, predictive models (classifying the intent of a user feedback message, for example) are fully supported, and OCR (extracting text from images) models are in beta. Currently, AppSheet’s OCR templates work on documents that have a fixed layout, and they require an internet connection.
Deploy your AppSheet application
Earlier it is mentioned that the user can view the apps on the supported devices by sending the users a link via email. This is called instant deployment. It is also possible to generate “white label” Android and iOS versions of its application and submit them to Google Play Store and Apple App Store for approval. Once approved, users can download them from the appropriate public store.
Overall, Google Cloud AppSheet offers a lot more features than it first appears. Yes, it is a no-code app builder, but the platform can also train and apply predictive ML models and OCR models.
Power Apps, Microsoft’s low-code app builder, is part of a larger ecosystem centered around Microsoft 365, OneDrive for Business, Power Automate, and Power BI. Google Cloud AppSheet is part of a similar but less complicated ecosystem centered around Gmail, Google Workspace, Google Drive, Google Maps, Google Sheets, Google Cloud SQL, and many Google Cloud services including machine learning, as well as Zapier. Amazon Honeycode, a slightly less capable no-code app builder, integrates with Amazon AppFlow and Zapier, as well as various AWS storage and data services. If a business already relies on Google for desktop functionality and/or data storage, AppSheet is an obvious way to produce custom apps to improve productivity.