Versions Compared
Key
- This line was added.
- This line was removed.
- Formatting was changed.
Soul Machines offers the following out-of-the-box Skills for you to leverage with minimal configuration in Digital DNA Studio.
| Info |
|---|
Currently, Skills are only created by Soul Machines and are vetted for potential risk to your projects. To support this each one is documented with as much transparency as possible. |
We categorize skills by their role in generating a response for a Digital Person to speak in response to user input. For detailed technical information see Skill Types.
Base Conversation
Base conversation skills provide sufficient functionality to drive an interaction by themselves. Examples include Sample Conversation and other monolithic corpora implemented using platforms such as Watson and Dialogflow ES/CX. When a base corpus skill is included in a project, all user input will be sent to it before any other default skill.
| Info |
|---|
Only one Base Corpus skill can be selected per project. |
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Fallback
A fallback type Skill is triggered when no other Skill can confidently match the intent of the user’s utterance.
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Intent Matcher
Unlike other skills, intent matching skills produce an intent classification. This classification can be used in multi-skill systems to direct user input to skills that will be more likely to provide useful output. For example, the input, “Do I need my umbrella today?”, might return the output “WEATHER”. This output is not intended to be sent to the user and is usually used internally to decide the next action to take.
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Pre-Process
Pre-process skills process user input before it is sent to any other skill (including base corpus skills).
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Post Process
Post-process skills process the generated response from the base corpus or default skills just before the Digital Person speaks it.
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Pre-post Process
This skill implements the functionality of both pre and post-processing skills in a single conversation turn.
| Info |
|---|
Adding more than one Pre-post Process type skill to a project may impede the performance of the skills added. It is recommended to use one of this skill type in a project. |
| Filter by label (Content by label) | ||||||
|---|---|---|---|---|---|---|
|
Why are some skills incompatible with each other?
Skills get triggered by specific User requests E.g. "What's the weather in Paris", should trigger the Weather Skill.
The same request could trigger multiple Skills. E.g. ‘What posts are trending today”, could trigger Facebook and Twitter skills. Thus these Skills cannot be used together as you can never tell which Skill might get triggered.
As the creator of a Digital Person, you need to decide which of these Skills best meets your needs and should be added to your project.
Best Practices
Try to limit the project to use only one PRE_POST skill, and one FALLBACK skill. It’s OK to have multiple Intent_matcher skills in one project.
Experiment and test before going live - check how Skills are triggered and the content they provide.
You can build a general use case with Skills alone, but it's still okay to make tailored conversations
Tailored conversations can still take advantage of Skills
Contents
| Table of Contents |
|---|