Trustwise, in collaboration with researchers from New York University’s Center for Data Science and Tandon School of Engineering, has introduced a research-backed AI optimization framework designed to improve performance, reduce costs, and enhance reliability for enterprise applications. The approach, outlined in a new technical paper, uses Bayesian optimization to strike a balance between speed, cost, and trust — key barriers in deploying scalable, high-stakes AI systems.
Matthew Barker, head of AI research at Trustwise, said their method helps enterprises “find the optimal balance” across competing objectives without relying on manual tuning. The system enables AI teams to optimize across multiple dimensions and deploy AI configurations that maintain accuracy while reducing operational latency and expenses.
The company also launched FinancialQA and MedicalQA, two domain-specific benchmarks for evaluating AI systems in real-world financial and healthcare scenarios. These benchmarks address the need for end-to-end evaluation in environments where response quality, real-time interpretation, and safety are critical.
NYU’s Umang Bhatt described the collaboration as a step toward “configuring AI systems to meet the unique demands of business,” rather than relying solely on model selection.
Trustwise’s Optimize:ai platform, now available for enterprise integration, allows organizations to operationalize this new framework and drive efficient, trustworthy, and compliant AI deployments. The company continues to support enterprises across regulated sectors with tools for AI safety, performance, and cost control.
