Telstra is increasingly relying on artificial intelligence to automatically detect and repair internet issues inside customers’ homes, as the telecommunications giant pushes deeper into large-scale automation aimed at reducing outages and cutting operating costs.

The system, known internally as SmartFix, has been operating behind the scenes for several years, remotely diagnosing broadband problems and rebooting customer modems before users even realise something is wrong.

While the technology lacks the public fanfare attached to many consumer-facing AI products, it has become one of Telstra’s largest practical deployments of artificial intelligence. During the 2025 financial year, the platform was resolving close to 7,000 network issues each day, helping the company avoid an estimated one million customer support calls.

Industry estimates suggest those avoided calls could be saving Telstra between $7 million and $15 million annually, with the figure continuing to rise as the system expands.

Telstra chief executive Vicki Brady has previously described the technology as giving the company’s broadband network a form of predictive maintenance capability, allowing services to “self-heal” before customers experience major disruptions.

The project originally began as a more limited machine-learning initiative focused on proactive network monitoring. According to Telstra’s group executive for product and technology, Kim Krogh Andersen, the system has since evolved into a far broader AI-driven platform capable of analysing enormous amounts of network telemetry gathered from customer modems and infrastructure.

The technology continuously monitors internet stability and connection quality, allowing it to predict when equipment is likely to fail or when signal issues are beginning to emerge.

Krogh Andersen said SmartFix handled more than 2.5 million faults last year and is expected to surpass that figure this year as the models improve.

The AI platform is also designed to adapt based on new problems it encounters. Krogh Andersen recently experienced internet disruptions at his own home caused by interference from a nearby Google Nest device. After identifying the issue and relocating the device, Telstra added the scenario into SmartFix’s growing library of troubleshooting responses.

The company says the system can also automatically shift customers from fixed NBN connections to mobile networks during outages to help keep households online while repairs take place. Customers are notified when these changes occur.

Not every fault can be solved remotely. Physical damage to fibre infrastructure or network hardware still requires technician visits, although the system can now automatically initiate some service requests when it determines on-site repairs are necessary.

Telstra believes artificial intelligence will eventually allow it to significantly reduce internet outages altogether by continually refining how problems are detected and addressed.

The growing role of AI is also reshaping how the company approaches its broader network operations. Rather than viewing telecommunications infrastructure purely as static assets, Telstra increasingly sees the network itself as an intelligent service capable of responding dynamically to customer needs.

Although Telstra insists AI is not replacing staff outright, the company’s latest financial results showed substantial cost reductions linked partly to productivity initiatives and lower labour expenses. For the six months ending December 31, 2025, Telstra reduced costs by $179 million compared with the previous year.

The company has invested heavily in AI partnerships to support its broader transformation strategy. SmartFix was originally developed through a collaboration with analytics company Quantium, while Telstra later committed $700 million to a second seven-year partnership with consulting firm Accenture focused on expanding AI capabilities across the business.

According to Krogh Andersen, SmartFix represents only one component of a much wider rollout. Telstra now has more than 400 AI use cases operating throughout the organisation, alongside several large cross-company automation projects.

One of the newest systems, known as CustomerIQ, launched earlier this year and uses artificial intelligence to predict why customers may be visiting Telstra stores by analysing complaint histories, loyalty scores and previous interactions.

Krogh Andersen said the real challenge for companies investing in AI is no longer building individual tools but scaling them in a financially sustainable way without allowing operating costs to spiral out of control.

He warned that businesses rushing into AI without strong operational foundations risk spending heavily without generating meaningful returns, effectively benefiting major technology providers more than themselves.