Beyond the model race, India targets the switchboard – Business News

Instead of competing with global technology companies to build foundational models, startups such as Indierouter AI, Staqu Technologies and NthEye are developing AI model routers that sit between users and multiple models. The router analyses a request and directs it to the model best suited to the task, while managing factors such as cost, speed, accuracy, reliability and data governance.

For instance, a routine customer query or email summary could be routed to a smaller and cheaper model, while a complex coding or reasoning task could be sent to a more powerful model. For sensitive information, the routing layer can also impose rules on which models can access particular data. This allows enterprises to use different models without having to manage each one separately.

High Capital Barriers

The shift is significant because building a foundational AI model from scratch requires massive computing capacity, large volumes of high-quality training data, specialised talent and substantial capital. “For most startups, competing directly with large global technology companies that have already invested billions of dollars in developing foundational models is therefore difficult,” Jaspreet Bindra, AI expert and co-founder and CEO of AI & Beyond, told Fe.

“Model routers are not necessarily competing with foundational models; they are helping businesses make better use of the growing number of models available to them,” he added.

Early Market Traction

The early signs of adoption are already visible in India. Audio-video analytics startup Staqu Technologies, which has incorporated model-routing capabilities into its agentic framework, has more than 100 clients using its AI Model Router and is targeting over 200 clients over the next year.

“The routing framework is already part of our real-world AI deployments, including facial recognition, number plate recognition, violence detection, attribute-based search and gait-based re-identification,” Atul Rai, CEO and co-founder, Staqu Technologies, said.

Indierouter AI, which opened its waitlist in July and entered open beta on August 20, has attracted about 150 developers. The company runs its own inference capacity in Hyderabad and is targeting developers and businesses seeking access to AI models on Indian infrastructure.

“We did not think India’s gap was a shortage of models. We chose the layer where the shortage was real: serving and routing open models on Indian GPUs, with billing and compliance built into the platform,” Bharat B, founder and architect, Indierouter AI, said.

NthEye has also integrated a tiered routing architecture into its core systems and is using it across high-stakes deployments involving five government clients. It is targeting 25 institutional clients over the next 12 months.

For Indian enterprises, routing could become particularly relevant because AI models can perform differently across languages, industries and use cases. It can also help businesses avoid using expensive models for every interaction and establish rules for handling sensitive information.

However, the opportunity may depend on whether these companies can build more than an API aggregator. “The real value will come from developing an intelligent orchestration platform that can understand the business requirement, select the right model, monitor performance, optimise costs and give enterprises greater visibility and control over their AI infrastructure,” Bindra said.