CrushBank adds new Generative AI capabilities to their knowledge management platform

CrushBank continues its evolution beyond an AI-powered search platform with the addition of new Generative AI capabilities to their platform.

Brian Mullaney, CrushBank’s CRO

CrushBank, which started out using IBM Watson to provide a more efficient AI-based search tool, and has since expanded to add Generative AI to their platform, has made some additional announcements about new things that they can do with data. They have announced several new features to their platform. They include new automated ticket classification and custom models, a Universal API and enhancements for Master MSP accounts.

CrushBank sells to MSPs, who actually control the customer.

“We don’t go to market with MSPs, but to them,” said Brian Mullaney, CrushBank’s CRO. “That’s where 90% plus of our revenues come from, although there are also some OEMs.

“We’ve been doing AI for five years,” Mullaney stated. “With the great increase in popularity around ChatGPT, some companies have slapped AI on the back of their domain and done some connections for ChatGPT, but that’s all they do. We are on our third version of our platform, and we are not turning over the keys to a Large Language Model. We talk about what we do with Narrow AI.”

CrushBank has significantly changed what they do with the AI. The company was originally focused on improving the efficiency of the IT Help Desk industry with an application which leverages the analytics of IBM Watson.

“Since then, we have rewritten the application to add functions to searches we have already done,” Mullaney said. “Right after COVID, we rebuilt the front end of the system to embed it inside a PSA system of record, so it can be looked at natively. We also automated the engagement between the two systems, to provide elegant integration where you are inside the system.”

Over the last 12 months, CrushBank has also added some major automations.

“With machine learning-enabled processes, we have gone from a search system to an AI powered knowledge system,” Mullaney noted. “It automatically classifies tickets based on time reading them. The market has always seen this as a weakness, but now we are able to do it at CrushBank, using a controlled custom model.”

CrushBank Budgeter is another relatively new automation.

“It takes all these fields which have always existed in PSAs, but which were never properly populated, and populates them,” Mullaney said. Another addition was CrushBank Insight, which uses Watson to glean emotion and sentiment analysis from help desk tickets, to provide information about how tickets are being handled.

The new announcements start with automated ticket classification and custom models. CrushBank had already let MSPs streamline ticketing systems through AI-powered auto-categorization and budgeting of ticket data. Now, it has improved its Ticket Classifier functionality to include the ability to see configured service boards and automatically configure improvements that previously required manual association. Users can now also create custom models to ensure even more accurate ticket resolutions.

“The big difference is the custom model aspect,” Mullaney said. “Before, it was impractical to allow every different organization to run their own model, because many organizations which are vertically focused might want to classify data differently. Before, we really only had two framework options. Now we can train custom models which really lock down how organizations bring in their own data.”

Mullaney said that this extension of automation is scheduled for September.

“It provides summarization across any levels of fields and time entries, and can let you go back at the unstructured data,” he indicated. “This wasn’t don’t before because of the tickets being unstructured data. But having a summary of this is very valuable. This shows how we have come a long way from being an AI powered search engine. We use Generative AI for this, but it is properly narrow and well defined.”

CrushBank is also now introducing Universal API, which removes the need to build a custom native API with every integration, so users can seamlessly integrate data from various systems, databases, and platforms into the CrushBank AI Engine.

“The heavy fit here has been building the native integrations via APIs,” Mullaney noted. “We have had about 17 of the most common. But the larger the organization, the greater the chance they have one that we don’t have an integration for. This will let them APIs you have into the customized Universal API.”

CrushBank is also rolling out enhancements for Master MSP accounts, giving users the ability to filter results based on an individual Child MSP and then subsequently, any of its clients when using Insight. This new structure also allows Child MSPs to share and see shared tickets with other Child MSPs, maximizing efficiency for support delivery.

“It creates another layer of accessibility and control beyond Master MSPs, where the Master sees all, and the Child appears as collections inside the master data lake,” Mullaney said. “The Master can also turn it off.”