We are excited to announce the release of the AI Prompt action step, now available in the blockstore. This feature empowers you to create custom prompts, including the use of variables. As a pro-coder you can create your own AI steps as well.
Good to know: This step is text to text. So if you want to summarise the answers of a form, you have to convert these to text first. But we’re planning to add multiple types, like text to object, object to text and text to array to make it easier to create AI use-cases.
In the example above, a project will be created in the first step. This project contains a long description. In the AI Prompt step a prompt is written to generate a short description from on the long description. In the last update step, the project will be updated with the generated short description.
Examples of what you can do with the AI Prompt agent step
Easily summarize text using the AI Prompt Agent.You can create prompts like:
Summarize the CONTENT
START CONTENT
BlogPost 144717775443 Use the power of AI in your actions
END CONTENT
Protect sensitive information by anonymizing text with the AI Prompt Agent. You can create prompts like:
Anonymize the CONTENT
START CONTENT
BlogPost 144717775443 Use the power of AI in your actions
END CONTENT
Improve the accuracy of your text by correcting spelling errors. You can create prompts like:
Correct spelling of this text: BlogPost 144717775443 Use the power of AI in your actions
Documentation
Read our documentation to learn more.
Up next
- AI Classifier Action Step (next week)
Stay tuned for the upcoming AI Classifier action step. With this feature, you can harness the power of AI to determine the content's topic and then create different action flows based on that analysis or use it to evaluating a prompt against multiple choices and selecting the most appropriate option. This is useful in decision-making applications where an AI must choose from several possible answers or actions.
- Convert document to text to be compatible with AI steps (december)
To be able to use the context of documents in your ai steps, you will need to convert it to text. We will provide a step that does this for you.
Soon you will have the ability to add AI search functionality to your datamodel and multi line text properties. This will give you the ability to create smart search functionality in your app. This will allow the enduser to ask full sentence questions instead of keywords to find data. For example, when a user is searching for ‘Rainforest’, it will also find documents about ‘Jungle’ but with a lower match score.
- Text to object, object to text, text to array (Q1)
The current step is text to text (as mentioned above), but we’re planning to add multiple types in Q1
That was all for now, your feedback allows us to continuously improve our product. Please let us know what you think about it. Happy building! 😃