Single Action allows anyone to easily utilize LLMs to create simple app services. Freedom of ideas and prompts make it simple to build the app you want, and it can be published anywhere with a public URL.
Single Actions are located on the left menu bar, and we will create a simple translation app as an example. To begin, click on the plus “+” button in the bottom left. This will create a new single action app.
We are given some basic information for the new single action app that has been created. We can see a preview of the app itself, the sharable URL generated, and the name of the app. Click on “Edit” as shown above to start making changes.
- Interactive preview of the app that is being created
- Title of the application
- Give a short description of what the app does. This will be visible below the title
- Start adding inputs to the app. These are going to be the user input needed to perform tasks and you can create up to 10 unique input fields, and there is no limit on the number of options that can be added for those inputs. In this example as mentioned above, we are creating a simple translation app.
- The first input added is the text to translate. We have tied that to a variable named “sentence” and the data to be captured will be a paragraph or short text.
- Now, we added a multiple choice input for different languages to be chosen from. A user can select more than one at a time.
- Finally, we added a single choice input for style. The user can only select one formality style for how they want the text to be translated
- Select the LLM model you would like to use.
- Create the prompt to utilize the data and have the LLM perform the desired task. Below the prompt window, we are given a list of variables that can be used based on the inputs created.
Side Note – Available inputs are:
- Short text – One or two lines of text
- Paragraph – Multiple lines and/or paragraphs of text
- Single choice – Choose one from a dropdown
- Multiple choice – Choose multiple from a dropdown
- Document (Basic) – Work with one uploaded or knowledge base document
- Document (Advanced) – Work with multiple uploaded or knowledge base documents
If we do require document input for our app, there are some options we must be aware of.
Document (Basic) – If we require one document to be included, we give the input a name and variable as shown with other inputs. It defaults to being a document upload.
If we want the user to be able to choose which document they want to work with that is already uploaded within the knowledge base, we can enable “Attach knowledge Base docs” toggle.
Document (Advanced) – If we require more than one document to compare or pull information from, we give the input a name and variable as shown with other inputs. It defaults to being a document upload as well. The difference compared to document (basic) is, we have the ability to add in prompting for map and reduce so that we can take the summaries of each document and prompt further.
- Setting Map Prompt in Document (Advanced) is a required option.
- The Reduce Prompt setting is not required and can be left blank and skipped if not filled out.
Similarly, if we want the user to be able to choose which document they want to work with that is already uploaded within the knowledge base, we can enable “Attach knowledge Base docs” toggle.
Now, back to the sample app. We get to test it out with the interactive preview!
- Type in the text to test the translation
- Choose the desired languages from the multiple choice input created
- Choose the style from the single choice input created
- Generate a response!
Review the output of the prompt including the variables. Make changes to the inputs or the prompt and regenerate a response if needed. Once we are happy with the results, we can click “Publish” to finish creating our single action app.
When running the app, any content created also has the ability to be downloaded.
Choose the document type to download the content.
