Creation Of an AI-PoweredClient Information Redaction Microservice for A Reputed Customer Call and Communication Services Company
Services
- LLMS
- Microservices
- Dashboard Optimisation
- Software Development
- DevOps
- Database Development & Management
About
While we cannot disclose the name of the reputed Customer Outreach and Call services company, they are one of the most influential names in this call services industry. They specialize in providing calling services to BPOs, Call Centres and other businesses dealing with customer outreach. They also provide their services to IT consultancy companies with client grievance departments and even marketing firms in the industry of customer outreach.
Background
This reputed company provides a lot of customer outreach services as a B2B business. Their clients usually subscribe to their call services because of how reliable they are. However, when the clients of this company subscribe to their services, they also need a lot of data in the form of prior call recordings for training their own customer executives. That is where this reputed call services company is faced with a challenge. They simply cannot share call recordings in their raw form because these call recordings contain sensitive client information. However, this call services company still needs to provide call recording data to their B2B clients for training purposes. That is where our paths cross, we being among the prominent names for custom LLMS AI solutions.
The Challenge
The challenge therefore was to create a microservices solution that would seamlessly convert recorded calls with sensitive client information into redacted versions of those recordings with omitted sections for sensitive information. Simply put, we had to create an AI LLMS solution capable of “bleeping out” sensitive information from call recordings like client name, address, financial information and any other sensitive specifics. Our solution had to be completely customizable with the reputed call services company being able to set the redaction parameters on what information to redact and what to keep. We also have to make sure our solution is integrated seamlessly within the existing systems used by the client.
The Objective
- Automated Call Redacting Microservices
- Customizable Redaction Parameters
- Service Integration with Existing System
- Very-High Rate of Accuracy
- High Rate of Audio Reaction Processing
The Solution
The solution that we came up with was exactly what the client envisioned, a seamless microservice that could be easily integrated with existing systems and that would enable high-rate of accuracy in the redaction of sensitive information from call recoding audio.
The Redaction Process
01
Call Recording Commences
The first stage of the redaction process begins when the customer executive ends the call with the client and the call recording has been created
02
System Is Initialized
After the call recording has been created, our redaction system gets initialized and begins the process of redacting and censoring all the sections of audio set within the custom perimeters by our client.
03
Redaction Procedure Initializes
The redaction process usually takes an average of under a minute within which you can do an unlimited number of reductions and set an unlimited number of parameters. This will simply mute those sections of sensitive information after it has been said by the B2C client to the executive.
04
Redaction Commences
Finally, we come to the final stage of the process where the redaction process is completed and that redacted recording file can be seen on the very simple-to-navigate dashboard of the microservice.
High Accuracy System
One of the highlights of our solution is that we made sure the system was highly accurate with results ranging between 95% accuracy to 100% accuracy all the time. The reason for this is simple and it has to do with security and the respect our client has for protecting the sensitive information of the B2C clients. While we cannot disclose the exact proprietary AI engineering techniques and tools we utilized for this solution, we can definitely tell you that this is a system with no bugs.
High Speed of Redaction
One of the other important things we needed to keep in mind for the solution was the fact that our reputed client records thousands of calls daily and they have a very high volume of operations. That is why we needed to make sure our system would work almost instantly making sure that our client never gets bottleneck by the performance of our microservice. That is exactly what we delivered with redaction durations taking an average processing time of 30 seconds.
The
Result
100%
Reaction Accuracy
210%
Training Recording Delivery
250%
Call-data Information Security
The reputed call services company finally has a robust solution that allows it to create training recordings from call recordings with redacted sections that contain sensitive information. This allows our clients to send demo and training recordings to their B2B clients and this helps their clients use the services with better efficiency.
Our full-proof AI-powered solution ensures sensitive client data never with call recording files and this ensures a safer customer outreach industry.Therefore, this improves the brand trust levels of our client, thereby ensuring more business for them.
This is a win-win situation for everyone, our client, their clients and the entire customer outreach industry.