Tech Mahindra Launches VerifAI, Gen AI Validation Solution For Enterprises
Through the solution, the company aims to help enterprises validate the end-to-end lifecycle of AI-based projects, enabling them to scale and speed up their AI initiatives.
Tech Mahindra Ltd., a provider of technology consulting and digital solutions, has launched TechM VerifAI, a solution for validating and verifying the outcomes of artificial intelligence and generative AI projects. Through the solution, the company aims to help enterprises validate the end-to-end lifecycle of AI-based projects, enabling them to scale and speed up their AI initiatives.
According to Tech Mahindra, VerifAI has a pre-built, 360-degree validation framework across the gen AI lifecycle, customisable metrics, and a microservices-based architecture that can be integrated into existing technology stacks of enterprises.
The solution will validate data quality in the discovery and pre-development stages and test AI models, frameworks, and hyper-parameters in the development stage to enable security and accuracy. It will also verify and tune the deployed models to make outputs consistent and explainable.
“Many companies have not moved from pilots and experiments to enterprise level adoption of AI, due to absence of a robust validation and assurance framework. TechM VerifAI addresses this need with a comprehensive framework for assessing, auditing, and certifying AI solutions across various domains and use cases,” said Kunal Purohit, president, Next Gen Services, Tech Mahindra.
Tech Mahindra will leverage its experience in validation and verification projects to scale AI initiatives. Implementing a full-stack AI solution includes elements ranging from infrastructure and datasets to rules, application programming interfaces, gen AI models, and visualisation layers. After the integration of TechM amplifAI offerings with assets and intellectual properties for AI deployment, the new solution aims to enhance AI assurance and enable timely verification and validation of projects.
The company also recently announced the launch of Project Indus, a foundational open-source large language model designed to converse in various Indic languages and dialects.
