Frequently Asks Questions
Xwhizz is developed as a support tool for agents, providing historically-supported editable suggestions at each dialog sequence. However, you can choose to have reliable suggestions be sent directly to the user or after a supervised timeout.
No. Xwhizz is predictive, using historical data he's observed to decide on the best next step. We believe generative AI is not ideal for typical corporate customer-service environments, as it introduces unnecessary problems such as hallucination and inability to control and fine-tune it's decision process.
Yes, as all Machine Learning AI, Xwhizz requires training. However, unlike other solutions, where training represents a huge pre-launch investment or whenever there are any business changes, Xwhizz trains by itself by observing the company's normal business as usual. So, zero human effort is needed even for the domain-specific training process.
Â
Xwhizz does not do Data Retention. It immediately transforms any data into irreversible number sequences called Tensors, storing only these numbers. Before that, all data is anonymised, so private information doesn't even reach Xwhizz. Furthermore, both the anonymiser and Xwhizz's core are prepared to work on-premises, even disconnected from the internet, so the information can be fully vaulted in-house.
No. Although any documentation can help accelerate Xwhizz's learning curve, you do not need to have any documentation nor to code it to any machine-friendly format. Xwhizz learns by example, gathering the most representative and efficient processes from the operators, as they perform their normal daily duties.
Yes. But, unlike other solutions, it doesn't have go through the learning process for every language. It uses a single pivot language, English, translating messages from and to the user's language. This enables Xwhizz to be much more efficient and learn faster, simplifying it's adoption in any new country.
Deploying Xwhizz is just a matter of calling a REST-based API, so it could be done in a matter of hours. The learning process, however, depends on the average amount of interactions and their scope similarity clustering. The most typical timelines we've experienced go from a couple of weeks to a couple of months. Of course, pre-existing historical interactions can significantly accelerate the learning process.
For Cloud-based solutions, you don't need additional infrastructure except any application integration needed to call the Xwhizz's REST-based API.
For on-premises installations, the required infrastructure needed for every environment may need to be studied by our specialists.Â
Yes, XWhizz is ready to be integrated with very little effort with all major CRM solutions.
Unlike other solutions, new content or processes do not require new documentation, new models or model training, unless you want to accelerate Xwhizz's normal learning process. Xwhizz will be indefinitely looking for statistically relevant changes from the existing company's processes, as these changes are adopted.
Xwhizz.flow is a self-generating graphical representation of the processes being observed and learned by Xwhizz. It's an important tool not only to manage Xwhizz, but to monitor and control the processes being adopted and followed by the team.
Yes, our services are available for any Xwhizz fine-tuning, learning acceleration, integration or any other type of proprietary development that may be needed. However, no service is required in the majority of scenarios, being Xwhizz self-sufficient on it's evolutive core service.
