How to create a new project? How do I make a new assistant?
If this is not your first project, you can start a new project by clicking on Create Assistant at the bottom of your dashboard lists.
When creating a new project, you will notice a modal window appear and you will be prompted to
- Name your assistant. This will not affect the assistant and can be changed later.
- Choose between two different types of assistants: “Launch & Host” and “Design & Handoff”.
- Select “Launch & Host” if you intend to publish the chatbot on existing chat and voice channels like a web chat modal and WhatsApp.
- Select “Design & Handoff” if you intend to develop the chatbot independently.
If you picked “Launch & Host”, you will see two dropdown menus for “Channel” and “Language”. There are two types of channels, which are chat and voice. Select a channel from either category depending on your needs, and whatever language you intend to design your chatbot with.
If you picked “Design & Handoff”, you will see three dropdown menus for “Modality”, “NLU”, and “Language”.
- For modality, you can select between chat and voice for your assistant.
- Natural Language Understanding (NLU) adds out of the box intelligence to your assistant (if this is a foreign concept, we recommend selecting the Voiceflow default option for the simplest user experience). You can learn more about NLU, or Natural Language Understanding systems that help interpret human language, from here
- For language, select the language you are designing, building and intending to deploy your conversation in
Important Note: Selecting your NLU allows you to build your assistant and leverage that platform/NLU provider's NLU Model, and DOES NOT connect directly with an NLU.
Each of the above channels will have access to different Steps, design/publish-aspects and features according to the native functionality of the channel you are designing for. For example, voice channels will have Speak steps while chat channels have Text steps, and only some chat channels will have Button steps.
Voiceflow supports several native integrations with major assistant platforms, but can also be configured to work with any custom assistant. If you have some technical skills and/or developer resources, you can leverage the Voiceflow Dialog Manager API to add any amount of custom functionality to Voiceflow that isn't supported natively.
NLU Selection (optional)
NLU systems work by analysing user input whether voice and/or text, and uses that to determine the meaning behind the user's request.
Once you select your channel/modality, you can choose your conversation's NLU compatible/available on Voiceflow today:
- Voiceflow (default)*
- Dialogflow ES
- IBM Watson
- Microsoft LUIS
- Salesforce Einstein
- Amazon Lex V1
- Nuance Mix
*If you don’t already use one of the NLU providers listed, we recommend selecting the Voiceflow default option for the simplest user experience. This will still allow you to export your designs into the production provider of your choice.
Note: Of the NLU types available in the list above, you are able to import any existing data with the respective NLU provider, as well as export in any data compatible with the respective provider in your Voiceflow project. To learn more about the import file types/formats supported, navigate to the NLU Content Imports doc here.
Once a custom NLU option is selected (ie. IBM Watson), you will also have the option to import a pre-existing NLU model file in the Create Assistant project creation screen. With NLU Data Import, it allows you to kick-start your designs using data exported directly from the NLU of your choice!
Note: Please note that toggling NLU to a custom provider listed above, is only available in our Voiceflow Enterprise plans. Please contact our sales team to learn more.
This can be seen with further options appear under this section/box, which indicates a link/button to upload the model, and a prompt of the file types supported with the import for the respective NLU.
Once the model import is complete, you will notice a notification/confirmation of the status of the import, such as the number of intents imported. You will also notice the section under the NLU selection box update with this data, and additional information such as number of entities imported.
Tip: Once project creation is completed & you are taken to the Canvas, navigate to the NLU Manager (shortcut M). You can verify and check that the NLU Model data imported successfully/correctly (ie. Intents, Utterances and/or Entities)