Bharti Airtel has successfully deployed an in-house Small Language AI Model across 30,000 field engineers. By shifting real-time image processing from cloud servers to endpoint smartphones, the telco has eliminated Rs 30–45 crore in cloud costs while raising field installation and repair standards.
Bharti Airtel, India’s second-largest telecommunications operator, has rolled out a patented, in-house Small Language Model (SLM) artificial intelligence tool to its network of 30,000 field engineers. Announced by top leadership during the company’s Q1 FY27 post-earnings call, the on-device AI system handles real-time image inference locally on standard smartphones, eliminating the need for expensive cloud processing.
The technological shift has allowed Airtel to reduce annual cloud workload expenditure for these field applications from Rs 30–45 crore down to effectively zero, while simultaneously improving workmanship quality, safety compliance, and fault resolution speed.
How On-Device AI Transforms Airtel Field Operations
Unlike traditional enterprise AI implementations that route image data to distant cloud servers for analysis, Airtel’s SLM operates entirely on the engineer’s endpoint mobile handset. The homegrown model performs computer vision tasks and standardizes operational workflows directly on site.
- Real-Time Processing: Field engineers receive immediate feedback on installation quality, optical fiber splicing, and physical cabling setups.
- Safety Adherence: The application verifies ground-level safety gear compliance before technicians begin hazardous tasks.
- Offline Functionality: Because inference happens locally on the handset, engineers can run diagnostic checks even in areas with weak cellular coverage.
Highlighting the engineering milestone during the analyst call, Shashwat Sharma, MD and CEO of Airtel India, noted:
“This quarter, we were able to make a meaningful difference in our workmanship across 30,000 field engineers with the help of AI. We are putting a small language model that works on the engineer’s endpoint device, which is a regular handset. This is a breakthrough for the engineering team, which we are using in-house.”
Slashing Cloud Workload Expenses to Zero
Prior to deploying the on-device SLM, processing image verification and diagnostic workloads required continuous cloud server computation. Transitioning these workloads to local smartphone chipsets has delivered substantial operational leverage.
Addressing the financial impact, Gopal Vittal, Executive Vice Chairman of Bharti Airtel, explained:
“We were spending, let’s say Rs 30–45 crore, on these workloads on the cloud. Now that it is being done on the device, the cost has gone down to zero. It is a very big breakthrough, and we feel that this can be extended to our retail stores as well.”
Comparative Analysis: On-Device SLM vs. Traditional Cloud AI
Airtel’s move highlights an emerging industry trend where edge computing replaces centralized cloud AI architectures to lower operational overhead.
| Parameter | On-Device SLM (Airtel Approach) | Traditional Cloud AI Architecture |
| Compute Location | Local smartphone hardware (Endpoint) | Remote cloud data centers |
| Annual Operating Cost | Near zero (post-deployment) | Rs 30 – 45 Crore per year |
| Latency / Response Time | Instantaneous (Real-time local processing) | Dependent on network latency and cloud load |
| Network Dependency | Works offline in remote locations | Requires active high-speed connectivity |
| Data Privacy & Security | Photos remain localized on device | Transmitted across external networks |
Operational and Financial Health in Q1 FY27
The deployment of cost-efficient AI tooling aligns with Bharti Airtel’s broader financial momentum in the first quarter of FY27. Supported by record postpaid subscriber growth and data consumption, Airtel reported:
- Net Profit: Exceeded Rs 8,000 crore for the quarter.
- India Revenue: Reached Rs 41,214 crore, reflecting a 9.7 percent year-on-year increase.
- ARPU Leadership: Mobile Average Revenue Per User rose to an industry-leading Rs 264.
What This Means for Consumers and the Telecom Sector
For retail and enterprise consumers, Airtel’s shift toward localized edge AI brings several practical benefits:
- Faster First-Time Resolution: Automated quality verification during initial installation reduces post-setup service visits and home broadband downtime.
- Structural Operating Savings: Lower operational expenditures help sustain capital expenditure for ongoing 5G and fiber network expansion without putting upward pressure on retail tariffs.
What AI technology is Airtel using for its field engineers?
Airtel has deployed a patented Small Language Model (SLM) that runs directly on the endpoint mobile smartphones of its field engineers.
How many field engineers are using Airtel’s small AI model?
Approximately 30,000 field engineers across India are actively utilizing the on-device AI system for daily operations.
How much money does the small AI model save Airtel?
By processing AI workloads locally on smartphones instead of the cloud, Airtel reduced its annual cloud computing expenses for these engineering tasks from Rs 30–45 crore to virtually zero.
What tasks does the small AI model perform on site?
The model performs real-time image analysis to standardize installation quality, guide fault repairs, and enforce safety measure compliance on the ground.
Will Airtel expand this AI technology to other business areas?
Yes, Airtel executives indicated plans to extend this on-device AI approach to retail stores and other customer touchpoints to further streamline operations.
