Trustwise and NYU Breakthrough Enhances Enterprise AI Reliability, Speed, and Cost-Efficiency

Addresses Enterprise AI Bottleneck to Optimize LLM and RAG Systems for Business ROI

AUSTIN, Texas--(BUSINESS WIRE)--Trustwise, the AI trust and optimization company, in collaboration with researchers from the New York University (NYU) Center for Data Science and Tandon School of Engineering, have unveiled a groundbreaking approach for optimizing AI performance for enterprises. By intelligently balancing cost, speed, and reliability, this new method enables businesses to deploy AI systems that operate more efficiently and effectively. Detailed in a technical paper, these advancements help better navigate trade-offs and compromises that have plagued enterprise AI initiatives.

Trustwise has also introduced FinancialQA and MedicalQA, two novel benchmarking datasets designed to evaluate end-to-end AI systems in high-stakes financial and healthcare environments, ensuring AI-driven decisions are both accurate and scalable. Enterprises today can implement the Trustwise Optimize:ai trust layer to operationalize this research-backed approach, enhancing AI efficiency, reducing costs, and improving response reliability.

Transforming AI Optimization for the Enterprise

The exponential growth of AI in business applications has created demand for more efficient and scalable optimization techniques. ​​Recent research reveals that 97% of data leaders face difficulty demonstrating AI’s business value due to competing priorities: reliability (43%), compliance (46%), and lack of trust (38%). Current AI tuning methods often prioritize a single objective, such as accuracy, at the expense of other crucial factors like reliability, speed, and cost.

For CIOs and CTOs, the advancements Trustwise and NYU’s Center for Data Science have identified represent a major step forward in enterprise AI strategy. Organizations in financial services, healthcare, and other highly regulated industries can now deploy AI models that optimize multiple objectives, including:

• Resource efficiency: Save weeks of manual AI tuning typically required for enterprise AI
• Risk mitigation: Prevent misaligned AI behaviors, harmful hallucinations, and data leaks
• Cost control: Reduce AI operating costs by up to 40% without sacrificing performance
• Performance optimization: Improve response speed while keeping accuracy and quality

Using Bayesian optimization, the research demonstrates significant improvements over baseline methods, providing enterprises with an optimal Pareto frontier — a set of configurations that balance multiple objectives simultaneously. By employing this methodology, organizations can deploy AI systems that maintain high-quality responses while minimizing latency and operational costs.

“Enterprise AI systems often force painful trade-offs between reliability, speed, cost, and carbon footprint,” said Matthew Barker, head of AI research and development at Trustwise. “Our approach gives businesses the ability to find the optimal balance for their specific needs, without requiring extensive manual tuning.”

Introducing FinancialQA and MedicalQA: New Benchmarks for Real-World AI Performance

As part of the research, Trustwise has also introduced FinancialQA and MedicalQA, the first benchmarking datasets specifically designed to evaluate AI performance in high-stakes enterprise environments. Unlike traditional AI benchmarks, these datasets reflect real-world constraints where AI must retrieve and interpret complex financial and medical information in real time.

“Optimizing AI performance is no longer just about choosing the best model,” said Umang Bhatt, assistant professor and faculty fellow at the NYU Center for Data Science and senior research associate at The Alan Turing Institute. “It’s about configuring AI systems to meet the unique demands of your business. Our approach provides enterprises with a systematic way to optimize AI for their specific needs, driving both efficiency and competitive advantage.”

Schedule a demo with Trustwise to learn how Optimize:ai drives successful production deployment and ensures reliable, aligned outcomes from high-stakes AI systems.

Additional Resources

• Read the Trustwise blog
• Follow Trustwise on LinkedIn and X

About Trustwise

Trustwise helps organizations innovate confidently and efficiently with generative AI. Its flagship product, Trustwise Optimize:ai, is a first of a kind generative AI application performance and risk management API that performs red-teaming and provides a robust AI safety, cost, and risk optimization layer for high-stakes enterprise environments. Trusted by enterprises across various highly regulated industries, Optimize:ai works with any AI model, supports various cloud, on-premises, and edge architectures, and is capable of handling large-scale generative AI operations and workloads. Founded in 2022 by a successful serial entrepreneur and the first general manager of IBM Watson, Trustwise is backed by leading investors and is headquartered in Austin, Texas, with research labs in Cambridge, UK, and New York.