Could your next AI advantage be sitting in your support tickets?
How Indian companies use small language models (SLMs) to turn domain knowledge into focused, affordable AI, with results from ManageEngine and NoBroker.
85% vs 60%
root-cause accuracy, SLM vs general model
Times Techies explores how Indian companies are using small language models (SLMs) to turn industry knowledge from years of support tickets and customer conversations into focused, affordable AI tools.
ManageEngine reports that internal tests of its SLM-and-knowledge-graph system identified the correct root cause of IT faults 85% of the time, compared with 60% for a similar-sized general-purpose model reading raw alerts.
At NoBroker.com, Akhil Gupta says shifting workloads to locally hosted models reduced costs by approximately 40–50%. Its conversation systems draw on Indian speech data to handle regional accents, background noise and switches between languages.
As CoRover founder Ankush Sabharwal puts it: "The objective is to match the model size and intelligence to the task."
Our takeaway
Start with a clearly defined workflow and the knowledge your team already holds. Test a focused model against real examples, measure accuracy and cost, and keep larger models available for tasks that need broader reasoning.
These are company-reported results, not guarantees for every deployment. They offer a useful reason to evaluate smaller models alongside larger ones.
Source: The Times of India, "India's Big Bet on Small AI", 30 September 2026 (Akhil George, Shelley Singh).
- Small Language Models
- Enterprise AI
- India AI