Airtel puts AI in the hands of 30,000 engineers. Here's why.
Why is Airtel putting AI on engineers' smartphones? The answer could change how businesses use AI at scale.
Not every AI model needs a massive data centre. Some fit in your pocket.
That's the idea behind Airtel's latest AI deployment. The telecom company has rolled out a homegrown small language model (SLM) to nearly 30,000 field engineers, allowing AI to run directly on their smartphones instead of sending every request to the cloud.
The goal is simple: improve installation quality, speed up fault repairs and strengthen safety checks, while showing how smaller AI models can solve real-world business problems.
What Airtel has built
Airtel's AI system runs on engineers' endpoint devices, including standard smartphones used during fieldwork. Unlike large AI models that require powerful cloud servers, a small language model is built for specific tasks and needs far less computing power.
That makes it suitable for devices that engineers already carry with them. According to Shashwat Sharma, Managing Director and CEO of Airtel India, the rollout has significantly improved the quality of work carried out by field teams. He described the initiative as a major breakthrough for the company's engineering operations during Airtel's Q1 FY27 earnings call.
Helping engineers in real time
The AI model uses real-time image processing to assist engineers while they work. For example, it can analyse images captured at an installation site and help verify whether equipment has been installed correctly or whether safety guidelines have been followed.
It can also support fault diagnosis, helping engineers resolve issues more consistently. The goal is not to replace engineers but to give them an AI assistant that improves accuracy and reduces rework.
Why Airtel moved AI away from the cloud
One of the biggest advantages of the rollout is lower operating costs. According to Bharti Airtel Executive Vice Chairman Gopal Vittal, the company previously spent between Rs 30 crore and Rs 45 crore on cloud infrastructure for these workloads. By shifting AI processing to engineers' devices, those recurring cloud costs have effectively been eliminated.
The AI models still need to be trained, updated and maintained. However, once deployed, much of the decision-making happens directly on the device. This approach also reduces delays because information no longer has to travel back and forth to a cloud server for every request.
Part of a larger AI strategy
The deployment is part of Airtel's wider push to use AI across its business. Right now, the firm is developing what it calls a homegrown agentic AI platform focused on automation, privacy, computing efficiency and data sovereignty.
Agentic AI refers to systems that can complete tasks with a degree of autonomy while operating within predefined rules. Airtel also plans to expand the technology beyond field engineers. Similar AI tools could eventually support employees in retail stores and other parts of the organisation.
What it means for the telecom industry
Telecom operators have increasingly adopted AI to improve network management and customer service. Airtel's latest deployment focuses on another area: improving the day-to-day work of employees in the field.
Instead of sending every request to the cloud, companies are beginning to run smaller, specialised AI models directly on phones, laptops and other edge devices. For businesses with thousands of employees, this can improve speed, reduce infrastructure costs and make AI easier to scale.
The bigger picture
Airtel's rollout shows that AI adoption is moving beyond experimental projects and into everyday business operations. By putting AI directly into the hands of 30,000 engineers, the company is demonstrating that practical, task-specific models can deliver measurable gains without relying entirely on cloud computing.
As more businesses explore on-device AI, Airtel's approach could become a blueprint for how organisations combine lower costs with better operational efficiency.