Got It, the company behind Excelchat, a human expert chat-based support service for Excel (and Google Sheets), is today introducing a new service called Querychat. The service both extends the Excelchat offering and seeks to compete with natural language BI solutions like ThoughtSpot, AnswerRocket, Arcadia Data’s Search, Tableau’s Ask Data and Microsoft Power BI’s Q&A.
Also read: Arcadia Data brings natural language query to the data lake
Information on Got It’s Web site, in its Querychat press release and from its Querychat explainer video can be distilled as follows:
- The support for Power Query and Power BI in Excelchat forms the basis of the service, allowing human experts to interpret customer-submitted natural language queries and generate corresponding SQL queries
- As a byproduct of the customer-expert interaction, a Natural Language Processing (NLP) model is trained using Google’s open-sourced BERT technology, in order to learn the customer’s jargon and conventions
- As the model becomes more comprehensive and better-trained, the natural language translation to SQL becomes automated, based on the incidentally trained BERT model. This facet of the service, called TrueNLP AI, is where it ostensibly becomes competitive with ThoughtSpot, Ask Data, Q&A, et al
- According to Got It, TrueNLP AI is superior to the aformentioned competing products, as the latter are based on keywords and synonyms whereas the Querychat model is based on semantic understanding of the NLP query terms
- Querychat will initially support data coming from Salesforce and Google Analytics, stored in Google BigQuery, with support for other SaaS applications and cloud data warehouses to be added
Mash-up derby for NLP BI
While Querychat could just be a blip on the BI radar, it is nonetheless intriguing, for a couple of reasons.
To begin with, it’s built on an unlikely mashup of Microsoft and Google technologies. Given Got It’s previous Microsoft stack pedigree, including Excel, Power Query and Power BI for Excelchat, it’s interesting that the company decided to build Querychat on BERT and BigQuery, rather than Azure Cognitive Services and Azure SQL Data Warehouse. Got It’s press release does say support for “Azure Cloud Data Warehouse [sic] and other environments” will be delivered this year, however.
Second, Querychat adds more critical mass and legitimacy to natural language BI as a category in its own right. While it remains to be seen how well the TrueNLP AI technology can function on its own, without intervention from Querychat human experts, the mere presence of another NLP BI player has significance of its own.
Regardless, it would seem the combination of human and machine intelligence might maximize results while keeping costs under control. And it will be interesting to see if any of the more established NLP BI players add sentence-based NLP to their current more keyword-based approaches.