Why Notable Capital is Excited about Patronus AI
May 22, 2024
As LLMs continue to revolutionize industries, enterprises are increasingly finding that the accuracy and reliability of their AI applications are critical to their generative AI success. Recent news stories have shown that deploying LLMs without robust evaluations can be embarrassing and legally damaging. That’s why we’re so excited to be leading Patronus AI's $17 million Series A along with Lightspeed Venture Partners, Datadog, Gokul Rajaram, Factorial Capital, and several leading software and AI executives. Patronus has quickly built the leading automated evaluation platform for LLMs, helping some of the world's largest enterprises deploy AI confidently.
Here’s why we believe Patronus is poised to become a cornerstone of the AI infrastructure stack.
The rapid adoption of generative AI has revealed significant challenges in maintaining the quality and safety of model outputs. Enterprises deploying AI risks hallucinations, personal information leaks, and copyright violations. Patronus is designed to mitigate these risks by evaluating model outputs in both development and production in real-time, ensuring errors are caught before they reach end users. This capability is crucial for enterprises aiming to get AI applications over the dependability hump, especially after high-profile cases of chatbot failures at companies like Google Gemini and Chevrolet.
Patronus' strength lies in its exceptional team. Co-founders Anand Kannappan and Rebecca Qian bring experience deploying models and ensuring quality at a massive scale. Anand helped Meta leverage AI to drive revenue and retention improvements within Meta Reality Labs before building an ML team at Vertis. Rebecca brings bleeding edge expertise from FAIR, where she trained and released FairBERTa, the first large language model trained with a fairness objective. This experience underpins Patronus’ independent approach to evaluating LLMs. In less than two years, Anand and Rebecca have established Patronus as both a leading research hub for AI alignment and a practical solution for one of the most pressing needs in enterprise AI.
Patronus has stood out with best-in-class proprietary evaluators to prevent hallucinations, PII leakage, and copyright violations, among other issues, allowing customers to see value from day one. The team has also built in the flexibility for enterprises to easily write custom evaluators to their specific use cases. Most impressively, we’re seeing customers take action with Patronus, blocking harmful outputs from reaching users and leveraging their data to build better models and AI apps.
The market for generative AI is exploding, and we believe every company that deploys LLMs needs Patronus. The company has already proven success working with everyone from the leading AI research companies to Fortune 500 enterprises.
By focusing on proprietary research and building custom evaluation models, Patronus is addressing the core problem in model quality – finding and preventing undesirable outcomes.
The AI infrastructure stack continues to evolve at warp-speed, but one challenge every team we speak with points to is the need for a trusted evaluation solution to ensure AI apps’ safety, compliance, and security. Patronus enables companies to confidently build and deploy reliable AI systems by detecting failures promptly and offering tailored evaluation solutions. As companies move from small AI pilots to full-fledged production deployment, we believe Patronus is a must-have.
Patronus is truly a notable addition to the AI infrastructure stack, and we couldn’t be more excited to support Anand, Rebecca, and the entire Patronus team as they build the market standard for AI evaluation. The team is hiring across the board to meet overwhelming customer demand - if you’re looking to get involved, check them out!
