Sarvam AI has introduced Prism, a new artificial intelligence decision model built on the company’s 105B-parameter foundation model. The release places the developer into the growing market for specialized AI systems designed to perform specific tasks rather than generate general text.
The Rise of Decision Models
Unlike traditional large language models that generate paragraphs of text, decision models are engineered to process inputs and provide specific, actionable outputs. These systems receive information and a defined set of potential answers, then return probability scores for each option. This methodology is intended to streamline processes such as customer support routing, content moderation, fraud detection, and security screening.
Expanding the Indian AI Landscape
Startups in India are actively developing alternatives to general-purpose AI systems to address local enterprise needs. ConvAI Innovations previously entered the space with Laya, an open-weight model designed for local device operation. Additionally, Sthānika AI, the research arm of Hyderabad-based PurpleTalk, created Sieve, a family of models that assigns probabilities to fixed answer sets. These companies argue that specialized models offer greater speed and efficiency compared to general-purpose systems.
Prism’s Technical Positioning
Sarvam AI states that Prism provides high accuracy while maintaining low response times. The model allows enterprises to customize performance and deploy the technology on their own infrastructure. The development reflects a broader strategic pivot within India’s tech ecosystem, where startups are exploring whether niche decision models offer a more viable path to compete with global AI players than attempting to match comprehensive, general-purpose capabilities.