AI services for comprehensive data support
At Microsoft, artificial intelligence from the cloud comes under the label of Cognitive Services. ASC uses these Cognitive Services are mostly in connection with compliance recording. This often includes conversations subject to European financial markets directives (Financial Instruments Directive, MiFID I / MiFID II). By compiling algorithms – i. e. Cognitive Services – artificial intelligence (AI) can be easily integrated into in-house services and applications. Users can have both audio and video data transcribed, translated, and analyzed. This is already quite common in today’s markets; what is rather new and exciting, though, is the next step: categorizing relevant and irrelevant information accordingly. For example, if keywords relating to security transactions pop up during a conversation, they will be identified automatically and will trigger subsequent measures. After all, not every conversation needs to be archived for the long term or assessed in detail.
Cognitive Services work in real time: The AI detects potential compliance risks as early as during the conversation itself which saves a lot of time. This way, employees no longer have to listen to the entire conversation all over again to gather relevant information and react accordingly. This process is based on transcription, i. e. speech-to-text conversion, which is particularly resource-intensive when done manually. Automatized structuring of the data, however, leads to increasing efficiency in the company. The goal is to transform non-searchable data into searchable and categorized data in a short period of time.
Successful implementation and use within Recording Insights
Recording Insights is the first ASC application which uses Cognitive Services. The pronounced advantage is that the recording solution tailored to Microsoft already exists in the same environment as the native app. This creates further benefits compared to previous solutions.
- No installation of other services or programs
- Flexibly scalable according to individual requirements
- No maintenance necessary
- No additional trainings
- Constant improvement of the Cognitive Services by Microsoft with the help of heuristic data
To ensure a low error rate in the results, users have to train the AI first, though. Correct training is essential to get a nearly error-free AI. Enriched with already categorized recordings and analytics, patterns are defined using Natural Language Understanding, i. e. comprehension of semantics and content. These patterns can then be applied to all following conversations. While Microsoft provides the foundation of neural networks as a service, users still have to adapt the system to their individual needs. Thus, services can differ in their efficiency even though they have the same foundation.
Data security and AI analytics
In the Microsoft Azure Cloud, geographic regions can be selected in order to adhere to national legal security standards. Therefore, Azure offers comprehensive options to guarantee reliability and to meet compliance requirements. With Cognitive Services, data can be processed nationally through the cloud or hybrid solutions. Consequently, users can ensure that the data does not leave its designated legal and economic area.
Independent of the demand in cognitive cervices, ASC offers advanced security standards. This includes security and surveillance tests, ISO certifications as well as continuous certification processes at Microsoft. All data is encrypted and protected against cyberattacks.
Natural dialog and further development
When it comes to compliance questions, Cognitive Services are still rather in their infancy. Users of AI are challenged with balancing costs against benefits. This specifically refers to how many contacts are potentially relevant and how high the costs for the AI will be until it can work independently. Here, the future presents the prospect of even more efficient systems which will require less time and effort to be productive – with regard to assessing data as well as to understanding conversations. Especially non-native speakers in a company will be able to reap benefits: Content, including the specific corporate wording, can be translated simultaneously with the support of AI.
Post-processing can be expected to take a major development step, too: summarizing conversations is already possible but does not yet render the ideal results for every use case. In the future, a consultation call may be transcribed automatically with the AI compiling a conclusive summary of the relevant information. Consultants no longer have to summarize their conversations themselves afterwards and thus have more time for high-quality customer service.
