Customer Service Automation · IT Operations Automation · Sales Automation
Amdocs needed to help telecom providers automate complex customer and network workflows with domain-specific AI while lowering inference cost and latency enough for near-real-time production use.
Reported impact
With NVIDIA NIM microservices and the surrounding telecom AI stack, Amdocs reduced token use by as much as 60 percent in preprocessing and up to 40 percent in inferencing, while cutting query latency by about 80 percent so end users receive near-real-time responses across commerce, care, and operations.
Amdocs built its amAIz platform on NVIDIA infrastructure to create telecom-specific generative AI agents for sales, billing, care, and network operations. NVIDIA DGX Cloud and NVIDIA AI Enterprise provide the training and enterprise AI foundation for commercially available and domain-adapted LLMs, while NVIDIA NIM microservices expose optimized inference endpoints and support frameworks such as NVIDIA TensorRT-LLM, vLLM, and SGLang. Amdocs also built an LLMOps pipeline around the NVIDIA AI Blueprint for building data flywheels, combining NVIDIA NeMo microservices for fine-tuning, evaluation, and guardrailing with NVIDIA NIM for deployment. On top of that stack, Amdocs uses NVIDIA Nemotron and Llama Nemotron to power amAIz agents that can autonomously handle multistep telecom journeys and plug into AI factories as reusable production services.
Telecom workflow context enters amAIz, DGX Cloud and NeMo-based data-flywheel pipelines adapt and guardrail domain models, NIM serves the resulting models and agents through optimized inference frameworks, and the agents return actions or responses for sales, billing, care, and network operations.
Amdocs built a robust LLMOps pipeline based on the NVIDIA AI Blueprint for building data flywheels; NVIDIA NeMo™ microservices for streamlined fine-tuning, evaluation, and guardrailing; and NVIDIA NIM™ microservices.
Amdocs is using NVIDIA DGX™ Cloud and NVIDIA AI Enterprise software to develop and deliver solutions based on commercially available large language models (LLMs) and domain-adapted models.
Amdocs reduced the number of tokens consumed for deployed use cases—by as much as 60 percent for data preprocessing and by up to 40 percent for inferencing.
The collaboration has successfully reduced query latency by approximately 80 percent, ensuring that end users experience near-real-time responses.
They’re also using NVIDIA NIM™, a set of easy-to-use inference microservices designed to accelerate the deployment of generative AI across enterprises.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
OpenAI Sample Jul 16, 2026 Cars24 Cars24 needed to manage a fragmented, conversation-heavy vehicle marketplace and its internal workstreams without scaling support, sales, and engineering headcount linearly with customer growth. 12% of previously lost seller leads recovered AI-powered voice and chat agents · ChatGPT Enterprise · Codex 
ServiceNow Sample Modified · Jul 30, 2026 Arqiva Arqiva needed to replace fragmented customer service and field operations workflows with a single platform that could coordinate customer journeys, schedule the right engineer faster, and give customers and staff one… 130% productivity improvement in Field Service Management CMDB · Customer Service Management · Field Service Management 
NVIDIA Sample Jun 26, 2026 MISA MISA needed to move from expensive, externally hosted LLM APIs to a self-hosted Vietnamese AI stack that could protect sensitive data, reduce latency and operating cost, and support many concurrent public-service and… $200,000 in monthly cost savings MISA AVA · NVIDIA A40 GPUs · NVIDIA DGX H200 