Our AI models are blueprints of intents trained with expressions. When creating an impact report, they are used to predict how messages can be grouped.
Over the years Ultimate has gained a vast knowledge of industry best practices when it comes to a well-functioning AI model. Based on this experience we have built a number of different industry models for you to utilize.
When creating an impact report, there are two options to choose from:
Choosing a model
In a nutshell, if this is your first time generating an impact report, have a look at the Industry Model table below: whenever there is a model available in your language and industry, use that in the impact report. If not, select Clustering.
However, if your bot has been running for a while and you want to gain insights on how you can improve your bot, Current Model would be the one to go here.
Industry Model
We offer different industry models specific to different languages. Our industry models help in automating your chat volumes and reducing your average handle time. The model generates a report that suggests topics found in your historical chat data based on our experience with the industry.
Current Industry Models
|
Industries / Languages |
Airlines | Gaming | Telecommunication | Banking | e-Commerce |
| EN | ✔️ | ✔️ | ✔️ | ✔️ | |
| DE | ✔️ | ✔️ | ✔️ | ||
| ES | ✔️ | ||||
| FI | ✔️ | ✔️ | ✔️ | ✔️ | |
| FR | ✔️ | ||||
| IT | ✔️ | ||||
| NL | ✔️ | ||||
| SE | ✔️ |
We're constantly working on improving these models and adding support for more languages and industries. If you don't see one of these options in your model selection, please let us know by submitting a request.
Current Model
Current Model allows you to compare your current AI model performance to a previous period by running your own existing data through your own model. This option is especially helpful for users who already have an Intent structure that has a good content coverage.
Once you have some data imported and are ready to go, go to Training Center > Impact Reports to generate a new report. To learn more, see Automation Report Explained
Clustering
Clustering is for when your bot is brand new and there is no industry benchmark model to be used for creating an impact report. When selected, messages in your historical data are grouped together based on their similarity automatically.