HomeNewsGlean desires to beat ChatGPT at its own game – within the...

Glean desires to beat ChatGPT at its own game – within the enterprise

GenAI has its problems. But if there's one thing it excels at, it's finding answers from massive pools of knowledge.

Input Read up, whose software connects to first- and third-party corporate databases to field plain-English queries (e.g., “How do I put money into our company's 401k?”) from employees, much like a custom ChatGPT. Launched by Arvind Jain, co-founder of cloud data management company Rubrik, Glean was inspired by Jain's observations that Rubrik employees often struggled to seek out the knowledge they needed to do their jobs – and that employees other firms were fighting it.

“I've seen engineers spend an excessive amount of time outside of code, account executives unable to seek out the most recent research or presentations needed to shut deals, recent hires taking too long to onboard, and so forth,” Jain said in a single Interview with TechCrunch. “This growing problem destroyed productivity, sapped energy and impacted the worker experience.”

It seems that Jain was on to something.

A current gardener Opinion poll found that 47% of desk employees have difficulty finding the information they should do their jobs. In the identical survey, employees reported that the growing variety of apps they should manage within the workplace – a mean of 11 today, up from six five years ago – is compounding the challenge.

In 2019, Jain – together with a small team of founders – built Glean as an AI-powered search app for enterprise customers.

The first few iterations were modeled after Microsoft's SharePoint Syntex and Amazon Kendra and occupied a product category generally known as “cognitive search.” Using natural language processing, early Glean could understand details about documents along with the searches that employees could perform across the corporate.

Photo credit: Read up

Over the years, Glean evolved right into a platform that connects to and analyzes an organization's databases and data stores to reply worker queries – following the explosive GenAI trend. Today, Glean captures information from sources like support tickets, chat messages, and entries on the shopper relationship management platform and applies GenAI to show all of it into insights and relevant answers.

One can imagine that firms could be wary of connecting their proprietary data – particularly their internal chat data – to a GenAI platform that performs such extensive scraping and evaluation. And that wouldn't be a unsuitable assumption.

A brand new Cisco Opinion poll found that multiple in 4 organizations have banned the usage of GenAI as a result of privacy and data security risks. In the survey, firms said they feared that GenAI tools would put their mental property in danger or potentially leak other sensitive information to the general public – or their competitors.

However, Jain claims that Glean is “secure” and “private” – a minimum of to the extent that it’s a cloud-based GenAI platform.

“Glean respects the identical permissions set across an organization’s data sources (Slack, Teams, Jira, ServiceNow, etc.), so employees only receive responses based on the information they’re allowed to access,” said Jain. “When a user deletes a document within the underlying application, the document is deleted from the Glean system.”

However, what concerning the curse that almost all GenAI suffer from – hallucinations? Is Glean protected from making up facts and quotes, getting summaries unsuitable, and missing the purpose of basic demands?

It is feasible; This creator was unable to check Glean himself. But while Jain declined to say how often Glean hallucinates, he emphasized the workarounds in place to make the platform's GenAI more reliable, including a model trained on customer data to learn industry and company-specific jargon, and the power for patrons to to change between multiple open source GenAI models to advance Glean's core experience.

“AI work assistants must deliver personalized results based on who’s searching,” Jain said. “Various elements of the searcher – their role, their job function, their management hierarchy, specific projects and responsibilities, and even who they work with – are ultimately necessary in defining the content that’s relevant to them. Glean learns a novel model for every customer to deliver highly personalized results to every worker based on those attributes.”

Glean also uses RAG (short for Retrieval-Augmented Generation), an increasingly common technique designed to “ground” GenAI by retrieving data from external knowledge sources to enhance performance. Jain says that every of Glean's answers “could be fully traced back to the unique source.”

“Collect (and recommend) documents that users might have for his or her every day work by learning from previous work patterns,” Jain said. “(It) delivers a turnkey implementation of a posh AI 'ecosystem' with over 100 connectors.”

Glean makes money by charging a monthly subscription per seat, based on annual contracts.

Despite competition from vendors like Microsoft (particularly Copilot) and OpenAI (ChatGPT), in addition to enterprise search vendors like Coveo, Sinequa, and Lucidworks, Jain says business has been quite good recently, with annual recurring revenue nearly quadrupling last yr.

Read up

Photo credit: Read up

This flies within the face of the narrative that firms – removed from fully embracing GenAI – have been slow and cautious in adopting it of their business functions.

Reply to a reply from December 2023 Opinion poll From Convrg.io, the Intel subsidiary, only 10% of firms said that they had brought GenAI solutions to production in 2023. The overwhelming majority of solutions are still within the research and testing phase, the businesses said – suggesting firms are pulling back from failing to seek out profitable GenAI use cases.

However, Glean's financials – and a 200-strong customer base that features Duolingo, Grammarly and Sony – appear to have won over investors.

Glean today announced that it has raised a Series D financing round co-led by Kleiner Perkins and Lightspeed Venture Partners, with participation from General Catalyst, Sequoia Capital, Adams Street, Coatue, ICONIQ, IVP, Latitude Capital and other strategic backers, Capital One, which has raised $200 million, includes Ventures, Citi Ventures, Databricks Ventures and Workday Ventures.

Kleiner Perkins' Mamoon Hamid said in an announcement: “The opportunity for Glean is gigantic and we’re so confident within the team's ability to deliver the GenAI solution to the corporate that we co-led this round after investing in each “round before, after managing Serie A in 2019. I've spent my enterprise profession investing in applications that enable knowledge employees to be more productive, be it Slack, Box or Figma, and see great potential in Glean to rework the best way people work. “

Jain says the brand new capital, which brings Glean's total revenue to about $360 million and values ​​the startup at $2.2 billion, will likely be put toward expanding “all” of Glean's teams (the one in Palo Alto based company currently has around 300 employees). Improving its product and “constructing a strong go-to-market movement.”

“Glean continues to see strong and growing customer demand, particularly from firms which have spent the last yr evaluating the vital requirements for adopting GenAI of their organizations,” said Jain. “We have at all times been prudent in our hiring and spending, and the recent increase in hiring is designed to satisfy strong customer demand.”


Please enter your comment!
Please enter your name here

Must Read