Agentic and autonomous AI for business excellence

Commercial context

In healthcare, there is a pressing challenge: how to support advanced AI imaging services at scale, without escalating infrastructure costs or straining clinical workflows. Hospitals increasingly rely on AI to speed up diagnostics and reduce error rates — from CT scan analysis to retinal screening — but these gains often require specialized GPU hardware, fast networks, and constant tuning. For many hospitals, especially in resource-limited settings, this level of investment is unrealistic.

CSPs, however, already manage distributed compute and network infrastructure. By orchestrating these resources intelligently, they can deliver AI tools as an on-demand service. That was the goal of the ‘Agentic and autonomous AI for business excellence’ Catalyst: to build a scalable, intent-driven operational environment that delivers AI workloads efficiently across telecom networks.

The solution

The Catalyst team built and demonstrated a cloud-edge platform tailored to AI medical imaging. The platform allows hospitals to access advanced imaging models through a telecom-delivered service, removing the need for in-house compute clusters.

Hospitals connect to the service via 5G. Diagnostic models are hosted centrally and deployed to local edge servers, where inference runs in real time. This ensures rapid, low-latency performance without compromising patient data privacy. The entire process — from service request to model delivery — is governed by intent-based orchestration.

CSPs manage this environment using several TM Forum assets, including the Business Process Framework (eTOM), Autonomous Network Reference Architectures, and APIs like TMF921 (Intent Management) and TMF640 (Service Activation and Configuration). These allow for automated deployment, monitoring, and scaling — with minimal manual input.

Operationally, this approach helps CSPs shift toward higher levels of autonomy . In the demonstration, the team showed that up to 60% of operational processes could be automated, resource utilization could increase by 30%, and operating costs could fall by 50%.

Wider application and value

The value for healthcare providers is immediate. One AI model used in the project analyzed a 3D CT scan in under 30 seconds. The same task typically takes a radiologist 20–30 minutes. Other models demonstrated a 37% drop in false positives during breast ultrasound analysis. Faster results, fewer errors, and reduced workloads mean better outcomes for patients and clinicians alike.

Cost savings are also significant. Hospitals no longer need to purchase and maintain costly GPU infrastructure. Industry benchmarks suggest switching to a shared AI platform can reduce total imaging costs by 50–70%. Because the service is centrally managed by the CSP, hospitals can be onboarded within days — no custom IT build-outs required. “We can automate 60% of operations, boost resource utilization by 30%, and cut operating costs in half,” said Yang Jianjian, Director at China Unicom, and Catalyst project lead.

The project also demonstrated broader advantages for the healthcare sector. A single telecom-hosted platform can serve hundreds of hospitals through the same infrastructure and model pipeline. This 'model-as-a-service' approach improves delivery times, increases consistency, and expands the customer base without linear cost growth. As AI medical imaging continues to grow — from $1.52 billion in 2023 to an estimated $6.35 billion in 2030 — scalable infrastructure will be key.

For CSPs, the benefits go beyond healthcare. The same architecture could support AI applications in logistics, gaming, or smart manufacturing. By coordinating compute and network operations across domains, CSPs can provide value-added services and position themselves as technology leaders. This creates new revenue streams and strengthens customer relationships.

The social benefits impact is clear too. Faster, more accurate diagnostics lead to earlier treatment, better outcomes, and lower costs. The WHO highlights that AI can close diagnostic gaps in underserved regions. In practical terms, this Catalyst’s solution could enable rural clinics to provide advanced screenings without local infrastructure. That means more patients screened, treated, and supported before conditions escalate. This Catalyst provides a clear demonstration — and a pathway — for how CSPs can power the future of AI in healthcare, bringing speed, scale, and autonomy to one of the sector’s most demanding and essential domains.