The AI race so far has largely been about models that can chat, code and reason. But a quieter shift is underway: decision models are emerging as an alternative to large models. As Indian startups and global giants enter the space, can decision models become AI’s next product layer?

What Are Decision Models? Unlike a conventional LLM, decision models do not generate prose. They return precise and structured outputs like probabilities and scores. Its approach is designed for bounded tasks and taking a decision – one that can be reviewed and governed later. General-purpose models can do this, but calling them at every step adds latency, cost and unpredictability.

The Growing Demand: Decision models are gaining more traction among enterprises as they are more affordable than LLMs. This is because they consume less tokens and bring strict output control, auditability and deep integration with proprietary internal data. While the momentum began with Jev, the first public model from TypeSafe AI, big giants like Cloudflare, Microsoft, AWS, OpenAI and Perplexity have also entered the space lately.

India Enters The Chat: Homegrown startups are also catering to this shift. PurpleTalk’s Sthānika lab recently launched Sieve, while ConvAI Innovations has also unveiled an open-weight model called Laya that developers can run locally on laptops or phones. Both target decisions rather than open-ended generation.

The Workflow Question: Industry insiders believe that the next commercial opportunity may lie in pairing decision models with proprietary data, feedback loops and a clearly defined customer problem. For India, the entry points include specialised models, routing infrastructure and applications built around high-volume enterprise workflows.

Source link