Custom Data Schemas allows you to create your own customized databases that both you and your AI Agent can interact with in order to perform its duties. They provide a key point of customization for your AI Agent.

Creating a new Custom Data Schema

Custom data schemas can be created by by going to the “Advanced” sub-menu of the “Configuration” menu.

You can then select “Create New Schema” in order to get started:

You should now see an empty form that allows you to configure your schema.

The first step is to assign your machine name. This fields gives a name for the schema that will be used to reference it, for example in URLs or in smart chain steps.

The machine name should be composed like a variable name would be in code. That means it should be lowercase and contain underscores instead of spaces. For example, if your data schema has the Title of “Sales Proposal”, then a good choice for the machine name would be sales_proposal.

The Title can be pretty much anything you want. But the Description field can be important because it will be provided to the AI if it ever needs to generate data in this format.

The key part of creating a data schema though is the Schema section which will be described next.

Adding Fields to your Schema

The Schema section is the key part of your custom data schema and will be where you spend most of your time. The Schema is defined using the JSON-schema standard, except with some additional fields and options that allow you to control the form of the data.

If you expand the options for the field, you will get something that looks like this.

Configuring Fields

Fields can have any of the following properties:

Property	Validation	Description
`type`	Must be `string`, `object`, `array`, `number`, or `boolean`	The type of data stored within this field. Matches the types available in JSON-schema
`title`	String	Title of the field. This is shown in the user interface
`multiline`	True/False	Whether or not the text-box used for this field will be multi-line. Only applicable for ‘string’ fields
`rows`	Number, 1 or greater	The number of text rows the widget should take up if its a multiline text widget
`show_in_table_view`	True/False	Whether or not this field should be available as a column when viewing the data table for this data
`table_view_order`	Number, 0 or greater	Indicates the order that columns will be disabled in the table view widget in the UI. Lower numbers get placed first, left to right
`description`	String	The description of the field. This is shown in placeholders on the form widgets, or in subtitles on groups with `type: object`
`placeholder`	String	Placeholder text on form inputs. Overrides the `description` field if provided
`form_input_type`	Must be `text`, `number`, `email`, `phone`, `address`, `url`, `full_name`, `company_name`, `checkboxes`, `datetime`, `currency`, `radio`, `dropdown`, `string_list`, or `json`	The type of widget that will be used on the user interface in order to edit this field.
`required`	True/False	Whether or not this field is required to have a value or is allowed to be left empty
`options`	String List	A list of strings that represent the values of options when using `dropdown` form input type
`optionLabels`	String List	A list of strings that represent the titles that will be shown in the user interface for the form input type
`default`	Any	The default value for this field

Selecting an Icon

If you choose to set your custom data schema as available in the left-side menu, then you need to choose an icon. You can do it by clicking here:

The icon search menu will then open up:

The database is very large and you can usually find an icon that suits your needs. Simply type in a term related to what you want and choose the style you want on the left.

Once you have selected the Icon and save, you will need to reload the page in order to see it show up on the left side menu.