After the beta test, our user research noticed that most users need to be guided and learn the best practices in order to build an efficient model. We decided to create a visual do & donāt to increase the success rate. After this feature was added, most users are now able to build an efficient model.
The challenge was to provide flexibility when using the tool, while at the same time adapting to any technical gaps that can arise from artificial intelligence. The experience also needed to be completely accessible. It was a challenge to find how to select and detect fields with just keyboard navigation.Ā
// Fields that aren't detected automaticallyĀ
To help the user get started, some fields and tables have been automatically detected. They're shown by dotted rectangles. To use those fields in their model, the user just has to click to select then validate the selection.
// Resizing fields that are automatically detected
If some fields in the document might not have been automatically detected by AI Builder.Ā
We allow the user to draw a rectangle around the field he isĀ interested in, give it a name, and validate the selection.
// Resizing fields that are automatically detected
If a field the user want to use was automatically detected, but doesn't contain all the values for that field, you can adjust the selection by selecting the field, resizing the selection, then validating the selection.
// Hiding fields
If the user having trouble selecting fields because the automatically detected field suggestions are incorrect, or there are fields too close to one another, he can click a selected or suggested field and hide it. To show all your hidden fields one again, select the Show hidden fields option that appears at the top.
// confirming fields
If a user draws new fields or resizes detected fields, they will need to confirm those fields in each of the documents that they had uploaded. In this step, you need to draw the fields that are marked as pending in the right panel to teach your AI model to recognize these fields for this type of document.
To draw a field on a document, the user starts by drawing a selection on the document then picking which field it corresponds to. They can also select Draw in document on the fields list to the right to draw a selection for a particular field.
To draw a field on a document, the user starts by drawing a selection on the document then picking which field it corresponds to. They can also select Draw in document on the fields list to the right to draw a selection for a particular field.
// Thanks toĀ
Mathias WEDLINGERĀ - Manager
Joel RICHEMAN - User Researcher
Julie COPLEY - UX Content Designer
Anand Mourouguessin. - Object Detection Designer
Aaron Monson - Prediction Designer
Joe Fernandes - PM
Mustapha LazrekĀ - PM
Julie COPLEY - UX Content Designer
Anand Mourouguessin. - Object Detection Designer
Aaron Monson - Prediction Designer
Joe Fernandes - PM
Mustapha LazrekĀ - PM
Josh Rennert - PM
My Dev team š¤
My Dev team š¤
