Cresta’s Spring Launch is packed with new features and improvements to drive customer value and make our users happy. Our team has been improving our products and shipping new functionality at a rapid pace. As a result, the list is so long that we couldn’t cover everything extensively in this series of blog post, but we would like to tell you about all of them. This blog post gives a summary of all new features and releases that are not covered in the other posts, but that also will help you drive exceptional results for your team.
Cresta now integrates with SnapEngage, Salesforce and Amazon Connect
The Cresta chat suite is now available on SnapEngage, Salesforce and Amazon Connect. With these, we support additional chat platforms with significant market shares. This unlocks fast and easy deployments and ensures that you can use Cresta on the chat platform you already have in place. In the next days we’ll launch with major customers using these new platforms.
Improved Knowledge Base suggestions
Accurately referencing a Knowledge Base to pull up relevant articles, is key to agents performing like experts. We have improved our technology which retrieves the relevant articles in an ongoing chat – by tuning it on historical chats for each customer. In this way, we’re now automatically learning from which articles top performing agents send to customers in the context of particular issues the customer is running into. This means that we have a deeper understanding of various questions which visitors have asked in the past and know better which articles are most relevant and effective to help resolve these questions, as demonstrated by top agents. We use this knowledge to predict in real-time which knowledge base articles are most relevant for a given conversation and in this way help agents resolve customer issues quicker and better.
More accurate Coaching Assist
Cresta’s real-time coaching assist feature is key for inculcating the required behaviors in agents. With the field of Artificial Intelligence fast evolving and new advancements coming out every day, we have internally revamped the way we build our Artificial Intelligence systems so that our engineers can easily experiment with and switch to the latest state-of-the-art algorithms in the field. Engineers can now try out new versions by changing only a couple of lines of code. This allow us to be more flexible going forward and ensure we are constantly using the best systems to provide more customer value. This means improved coaching assist for all customers, with sustained top of the class accuracy going forward.
Streamlined Suggestions Training and QA
Cresta’s contact center agents get real-time suggestions on how they can respond to their customers. These suggestions make agents more efficient and help agents formulate the right response. The key to quality suggestions are models trained on up-to-date data. We built internal tools to streamline training and QA of our models to keep them fresh. Training and QA is now in a single portal where multiple people can collaborate and there is a single source of truth for our models and QA results. This allows quicker iterations on our models, ultimately providing better suggestions to agents faster.
Faster Time-to-Value
Cresta provides direct value to your customers so we’ve been working hard to reduce the time from signature to seeing business results. We streamlined the workflows in our internal machine learning platform, and have built new state-of-the-art algorithms that allow us to deliver models that require significantly less training data without sacrificing on accuracy. Finally, we have also improved our deployment pipelines, so that new models can now be tested and deployed to customers with the single click of a button. These improvements mean that new customers can now be completely set up and onboarded faster than ever before. With most customers, we’re seeing KPI results in just 15 business days from signature!
Scaling Customer Monitoring
Cresta provides live fully automatic hints to help agents navigate conversations and execute on key behaviors. We have developed custom tools to internally monitor hint quality and volume for each customer. It allows us to configure custom alerting that notifies us if any issue with hints is detected, allowing us to respond faster. In addition, the new tooling provides us with a better toolbox to find the root cause of an issue. All of these together mean we can now more quickly spot and fix any improvement points across any of our customers to make hints always appear at the right moment in the chat for agents to succeed and ultimately drive KPIs.
Special thanks to all the squads at Cresta Engineering who made this possible.