GenAI has a problem. But if there's one thing this product does well, it's the ability to derive answers from a vast pool of data.
Glean's software connects to a company's first-party and third-party databases and sends plain English requests from employees, such as custom ChatGPTs (e.g., “How can I invest in our 401k? ” etc.). Founded by Arvind Jain, co-founder of cloud data management company Rubrik, Glean was founded by employees at Rubrik who often struggled to find the information they needed to do their jobs, and employees at other companies as well. It was inspired by Jain's observation that he was struggling.
“We see engineers spending too much time on things other than code, account managers not finding the latest research and presentations they need to close deals, and onboarding new employees taking too long. “We did it,” Jain said in an interview with TechCrunch. “This growing problem has destroyed productivity, sapping energy, and hurting the employee experience.”
Jane seems to have noticed something.
A recent Gartner study found that 47% of desk workers struggle to find the data they need to get their jobs done. In the same survey, workers said they now have an increasing number of apps to manage at work (an average of 11 today, compared to an average of six five years ago), making the challenge even worse. It is reported that.
In 2019, Jain and a small founding team built Glean as an AI-powered search app for enterprise customers.
The first few iterations were along the lines of Microsoft's SharePoint Syntex and Amazon Kendra, occupying a product category known as “cognitive search.” In its early days, Glean used natural language processing to understand document details in addition to searches that employees across the organization might perform.
Image credits: gleaning
Following the exploding GenAI trend, Glean has evolved over the years into a platform that connects and analyzes enterprise databases and data stores to answer employee queries. Glean is currently ingesting information from sources such as support tickets, chat messages, and customer relationship management platform entries, and applying GenAI to transform it all into insights and relevant answers.
We imagine that companies will be wary of connecting their proprietary data, especially internal chat data, to a GenAI platform that performs such deep levels of scraping and analysis. And that's not a false assumption.
A recent Cisco poll found that more than one in four organizations prohibits the use of GenAI due to privacy and data security risks. In the poll, companies said they were concerned that GenAI tools could infringe on their intellectual property or leak other sensitive information to the public or their competitors.
But Jain insists that Glean is “secure” and “private,” at least as a cloud-based GenAI platform. can Get used to it.
“Glean respects the same permissions set on corporate data sources (Slack, Teams, Jira, ServiceNow, etc.), so employees only receive answers based on the data they are authorized to access. ” said Jain. “When a user deletes a document in the underlying application, the document is deleted from her Glean system.”
But what about the curse that most GenAIs suffer from: hallucinations? Is Glenn immune to making up facts and quotes, getting his summaries wrong, and missing the point of basic requirements? ?
It's possible; I haven't been able to test Glean myself. However, while Jain neither confirmed nor denied Glenn's hallucinations, he highlighted the mitigation measures in place to make the platform's GenAI more reliable. This includes models trained on customer data to learn industry and company-specific terminology, allowing customers to switch between multiple open source GenAI models to drive his Glean core experience. can.
“AI task assistants should provide personalized results based on the searcher,” says Jain. “Various aspects of the searcher are important in defining content that is relevant to the searcher, including their role, job function, management hierarchy, specific projects and responsibilities, and even the people they work with. Glean learns a custom model for each customer and delivers highly personalized results to all employees based on these attributes.”
Glean also employs RAG, which stands for Retrieval-Augmented Generation. This is an increasingly common technique used to “ground” GenAI by retrieving data from external knowledge sources to improve performance. Jain said all answers provided by Glean are “fully referential” to the original source.
“Gleaning” [can recommend the] By learning from past work patterns, we can create documents that users may need for their daily work,” says Jain. “[It] Provides turnkey implementation of complex AI 'ecosystems' with over 100 connectors. ”
Glean makes money by charging a per-seat monthly subscription based on an annual agreement.
Despite competition from vendors such as Microsoft (particularly Copilot) and OpenAI (ChatGPT), as well as enterprise search providers such as Coveo, Sinequa and Lucidworks, Jain says business has been very strong recently, with annual recurring revenue has nearly quadrupled. last year.
Image credits: gleaning
This goes against the narrative that, far from embracing GenAI wholeheartedly, companies are slow and cautious in implementing it across their business functions.
A December 2023 study by Convrg.io, an Intel subsidiary, found that only 10% of organizations said they had deployed a GenAI solution into production in 2023. Most of the solutions are still in the research and testing phase, the organization says – implying that companies have not been successful in finding profitable GenAI use cases.
But Glean's financials and customer base of more than 200 companies, including Duolingo, Grammarly, and Sony, appear to be appealing to investors.
Glean announced today that it is co-led by Kleiner Perkins and Lightspeed Venture Partners, with participation from General Catalyst, Sequoia Capital, Adams Street, Coatue, ICONIQ, IVP, Latitude Capital, and additional strategic backer Capital One Ventures. Announced that it has raised $200 in Series D funding round. , Citi Ventures, Databricks Ventures, Workday Ventures.
Kleiner Perkins' Mamoon Hamid said in a statement: “I co-led this round after making all the investments because the opportunity for Glean is huge and we have great confidence in the team’s ability to deliver GenAI solutions to enterprises. Having spent most of my venture career investing in applications that make knowledge workers more productive, like Slack, Box, and Figma, I see great potential for Glean to change the way people work. That means.”
Jain said the new funding, which brings Glean's total funding to $358 million and values the startup at $2.2 billion, will benefit “all” Glean's team (the San Francisco-based company has The funds will be used to expand the number of employees (currently approximately 300 people). Develop that product and “build a strong go-to-market movement.”
“At Glean, we continue to see strong and growing customer demand, particularly from enterprises that have spent the past year evaluating the requirements needed to implement GenAI into their organizations.” Jain says Mr. “We have always been prudent in our hiring and spending, and our recent hiring increases are in response to strong customer demand.”