Sunnyvale-based organization Brevian aims to streamline the process by which enterprise users can develop bespoke AI agents. At present, the organization is centering its efforts on security analysts and support teams, domains in which the training modules and use cases are precisely delineated. Nevertheless, there are plans to gradually broaden its scope to encompass additional areas. Brevian is emerging from stealth mode and announcing a $9 million seed funding round today, the company exclusively told TechCrunch.
BREV/νN, an acronym for Brevian, was established by Vinay Wagh (CEO) and Ram Swaminathan (CTO), two individuals who approached this entrepreneurial endeavor from contrasting perspectives. Wagh formerly held the position of product director at Databricks and served as the chief of products for Bracket Computing.
He oversaw the preparation of Databricks for sale to large enterprises and the subsequent release of its serverless product, Databricks SQL Serverless, during his tenure at the organization. “That went official in February, and I was observing the situation,” Wagh explained.
“Upon encountering ChatGPT and other similar technologies, my initial thought was: How can this be implemented in an enterprise setting?” Our immediate consensus was that each of these lacking components would be indispensable. Despite its considerable potential, its effectiveness will be compromised unless this component is resolved. That sparked the motivation to venture forth; significant events are transpiring. Following that, I became acquainted with Ram [Swaminathan] via a mutual acquaintance.
Although Wagh is primarily focused on data, Swaminathan possesses a foundation in computer science and machine learning. Following a period of employment in academia, he devoted nine years to Bell Labs and seventeen years to HP Labs, where he investigated coding theory, cryptography, and early machine learning initiatives, among other things. Subsequently, he assumed leadership of the AI trust team at LinkedIn until his departure in late 2022 to assist Wagh in the establishment of Brevian as its CTO.
Brevian founders Vinay Wagh (l.) and Ram Swaminathan (r.). Image Credits: Brevian
“In all honesty, I lacked product sense,” he explained. “You write papers and patent applications, but are you truly capable of developing a product that can affect a large number of people?”
And LinkedIn was an ideal target: by visiting, you contribute to the security of a platform that facilitates employment and personal connections. Such is it. It has significantly influenced society. “I derive great satisfaction from collaborating with engineering and product personnel to effectively coordinate these endeavors and generate substantial value.”
Security is the foundation of all enterprise initiatives involving generative AI, according to the team. For instance, AI providers must guarantee that no personally identifiable information escapes. At the time of the product’s introduction, numerous businesses decided to prohibit ChatGPT within their operations because it was impossible to ensure that employees would not input sensitive information into the chat prompts.
Early on, Brevian was preoccupied primarily with security. “Models that identify personally identifiable information were developed expeditiously,” Wagh elaborated. “In response to the growing prevalence of this prompt injection attack vector, we developed an intent-based prompt injection detection system.” As a result, our approach at that juncture involved amassing every prompt that violated alignment, executing a model across them all, and successfully identifying each one.
Image Credits:Â Brevian
Wagh observed that despite the fact that businesses continue to be extremely concerned about security in regards to large language models, they are unsure of the precise nature of their concerns beyond these breaches. The team rapidly came to the realization that the primary obstacle impeding enterprise adoption was not security per se but rather the development of systems that effectively addressed practical challenges faced by the organization.
“Our objective was to empower enterprise business users to utilize AI to streamline their routine responsibilities.” Everything aligns with the overarching goal of progressing beyond the realm of security and developing these artificial intelligence agents.
Brevian’s seed round was directed by Jake Storm, a partner of Felicis; this is the largest seed round check to date for Felicis. He observed that while there was widespread discussion regarding AI infrastructure a year ago, there was relatively little attention paid to applications and the criticality of securing them. “Slightly, 2023 was the year of AI infrastructure.” “AI applications will dominate in 2024, and we felt they were so far ahead of schedule,” he explained.
Brevian intends to accelerate product development and expand its team with the proceeds of this funding round in order to satisfy the rising demand of its early release program’s customers.
