What's new
This section contains release information for the eva server Conversational AI 3.4.1. All information in this section describes the last updates for the last version published.
Last updated
This section contains release information for the eva server Conversational AI 3.4.1. All information in this section describes the last updates for the last version published.
Last updated
The following features and improvements were developed for the current version: 3.4.1 - GA.
Ability to send the Intent directly to the eva CAI avoiding having to call the NLP engine and using it to resolve the answer. It is an interesting feature if you are integrating external tools or channels which are using their own NLP.
Now is possible to send the Project and Channel information to the masking module to improve the logic of how the data is obfuscated. We also have included support for custom Entity masking.
Until now, there was only a single default data region for all Dialogflow projects, located in the United States. With this enhancement, Dialogflow offers three new data regions: London, for all of Western Europe; Tokyo, for Northeast Asia; and Sydney for Southeast Australia. For the Americas and all other geographic areas, the previous default U.S. region will remain. The possibility of choosing a data region physically close to where the project we are developing in Dialogflow is going to be used has two main advantages:
Better responsiveness. As there is a less physical distance between the clients and where the data they consult is stored, the latency decreases, improving the response speed of our projects. For example, if the users of our virtual assistant are located in the Iberian Peninsula, and we select the europe-west2 data region, the latency will be more favorable.
Legal compliance. As stated by Google in its release note informing about this new feature, the choice of data region will facilitate compliance with regulatory requirements and privacy policies governing the projects we develop in Dialogflow. On the other hand, the choice of the region will not affect the cost applied to our project, since the price is the same for all regions.
Intent and Entity information returned by the NLP engine has been included in the USER INTERACTION table.
Communication with HTTP services has been improved by using webclient instead of Feign. Feign is kept for static URLs and webclient is used for dynamic URLs.
A new session timeout process has been included with improved performance in high load processes and is completely unlinked to the broker.
Now we provide Helm charts to facilitate the installation of the eva server.
, a new feature that works with information available in eva’s system that doesn’t depend on APIs
The new feature allows you to manage and customize the flows according to the variations of the business rule. It’s a resource to make your dialog more assertive and much more precise, as the virtual agent will respond to any changeable scenario
: Now, with eva, it is possible to predict in your flow different responses for the same Intent
: a template with 38 ready-made flows fitting in 10 banking use cases that you can customize as you like. Designed with eva and market best practices, so you and your team can optimize time and processes when building a virtual agent.