An organisation’s widespread commercial rollout of LLM-based products is often hindered by a lack of transparency and predictability, including such unexpected outcomes as factual errors, hallucinations, biases, or other deviations from the designed behaviour. All of which can result in great financial and reputational harm to the companies that rely on them. To help its clients mitigate these risks in real time, Verax AI also announced the launch of its enterprise-grade software solution, Verax Control Center. The state-of-the-art solution is designed to provide visibility and control of LLM-based products in production, enabling trustworthy and responsible AI.

Verax AI was founded by Leonid Feinberg and Oren Gev, who collectively have more than 45 years of experience developing enterprise-grade software solutions and scaling technology companies. The two first began working together in 2017 at CloudEndure, a company co-founded by Feinberg and where he was the VP of Product, which was later acquired by Amazon Web Services (AWS). The two continued working together at AWS until 2023, when they co-founded Verax AI together.

Having exited stealth mode, the company plans to use this fresh capital to accelerate its go-to-market processes and increase awareness. 

“We are excited to be able to simultaneously announce our Seed round and the release Verax Control Center. Providing customers with increased AI trust and a new level of visibility into how their AI products behave is a crucial step in protecting against risk and facilitating the widespread adoption of responsible AI,” said Verax AI Co-Founder and CEO Leonid Feinberg. “We’re seeing many parallels between the challenges organisations are facing when adopting AI today and the ones they faced adopting the public cloud a decade ago. As a result, we are perfectly positioned to help them.”  

While traditional Machine Learning (ML) Dev tools and Ops tools help optimise AI products at development and deployment times, Verax AI is the first to successfully mitigate risks once these AI products are moved to production. Its solutions provide visibility into the production use of LLM-based products outside the lab and auto-correction of unwanted behaviour in real-time without time-consuming configurations or human work.

“Companies of all sizes, across all industries, need to mitigate the risk of their production AI applications providing inaccurate, offensive, or embarrassing answers,” said Schuster Tanger, Co-Managing Partner of TQ Ventures. “Successful entrepreneurs with years of experience developing and deploying enterprise-grade solutions, Leo and Oren have developed solutions that are uniquely qualified to meet the significant challenges faced by these companies.”

For more startup news, check out the other articles on the website, and subscribe to the magazine for free. Listen to The Cereal Entrepreneur podcast for more interviews with entrepreneurs and big-hitters in the startup ecosystem.