Egnyte introduced some new generative AI tools today, and revealed more of what’s coming on the AI roadmap. The company also pointed out that it has been incorporating AI into the platform for over a decade, and is not simply jumping on the bandwagon of the moment.
Today, it’s announcing a ChatGPT-like interface where customers can interact with the content stored in Egnyte to do things like create summaries of documents or review transcripts of video and audio to generate a list of key points or summaries.
The company isn’t new to AI, but in the past, as with many enterprise companies announcing generative AI this year, it was taking care of background tasks like privacy, security and infrastructure management. Today’s announcement brings AI to the forefront to any business user who wants to interact with the content stored in Egnyte.
“We’re opening up the intelligence engine for every single user on the platform, which includes every business user from all departments. What that will enable them to do is through a chat-based interface, they will be able to ask and answer questions,” David Spitz, chief strategy officer at Egnyte, told TechCrunch.
In the future, the company plans to allow customers to query across a set of documents in a single folder or across multiple folders to locate content related to a particular subject, and that will bring powerful search capabilities to the content repository.
“Later this year, we’re introducing the ability to ask questions across sets of documents, which is like next generation search. You can think of it like natural language-based search,” Spitz said. This could help users find photos with a particular object in an image library, or summarize the last four years of drug discovery for a life sciences application.
Egnyte co-founder and CEO Vineet Jain says they are building their solutions on existing foundational models. Today, the models are built on GPT-3.5 and GPT-4 via Microsoft Azure, but he says they still need to fine tune these models for the specific requirements of their customers, who often have very specific language in areas like life sciences and financial services.
“The new products use the large language models from Azure [and OpenAI] – GPT-3.5 and GPT-4 – to which we have added proprietary mechanisms to verify that the results are accurate and appropriate and to prevent hallucinations. We also worked closely with Microsoft to ensure that no customer data could be shared with the foundational models or other customers, and to ensure that any information generated by the models would respect the access controls of the customer,” Jain said.
While today’s announcement is in line with what many enterprise companies are doing, it is essential for Egnyte to have its own spin on it to remain competitive. Box announced generative AI capabilities will be coming to the platform this year in May, for example.
Egnyte has never been one to be pushed by the tech of the moment, taking its time and building up a business while competitors took bigger chunks of investment capital and tried to move faster. The company, which launched in 2013, surpassed $200 million in revenue in the first quarter this year with revenue growth of around 25%.
Like other enterprise companies announcing these capabilities, Egnyte is offering the new tooling to a limited number of customers for now while tweaking it in production environments. It’s vague on when it will be more generally available, but probably some time later this year.