A client of ours had recently introduced a new workflow and reporting process which allowed them to manage client interactions very efficiently. The process has been optimised through years of hard-earned experience and was unique to their particular business.
The problem they faced was that the new process required the application of tags and decision points to every customer interaction – they had 25 years of history and millions of free-text records containing valuable information that was completely incompatible with the new process – effectively their past was a black hole.
Valiantly, an internal team had been drafted to manually work back through the historical records retro-fitting tags against previous interactions. This would enable them to fit historical records to the current process, providing valuable insights and surfacing important trends. However, they soon realised that even working through a few thousand records a day, this task would take years to complete and cost a small fortune.
Enter Robosoup – we specialise in prediction algorithms, information extraction and multimedia classification. Using the work that the tagging team had accomplished so far, we were able to build and train a classification system to quickly and accurately work through the remaining several million free-text records.
The core of the solution was a deep convolutional neural network built using TensorFlow. The new algorithm was able to compensate for spelling and formatting errors and navigate some gnarly business rules. It can make concrete decisions, where previously these had been highly subjective. Once complete, the client was given the source code for everything we developed, allowing them to freely modify and extend going forward.
The ability to harness more data, from more sources, in less time, and empowering users to collaborate and analyse data in different ways leads to better, faster decision making. With improved data aggregation, AI models and predictive analytics, your company can provide smarter and quicker customer journeys that build long-term trust and loyalty.
The new classification engine is now part of the current operational process and acts as a watchful helper in the day-to-day running of the business.
Contact us if you would like to learn more about using AI and machine learning in your business.
Or, follow us on Twitter for more updates like this.