Breaking

NVIDIA-powered offline medical AI CMES reaches 17,000 workers in 19 countries

July 7, 2026 at 14:04 EDT

Healthcare
AI Agents
Infra & Chips

CMES (CMES-AI), an offline AI medical-support platform built by the nonprofit Techies Without Borders (TWB) using NVIDIA technology, has been deployed to more than 17,000 healthcare workers across 19 countries and is helping improve care for 20 million people a year, NVIDIA said.

A Doctor's Assistant in a Box

Offline AI puts a "second doctor" in the world's hardest-to-reach clinics

CMES runs entirely on local edge hardware — no internet required — returning evidence-based answers in local languages, and now improves care for 20 million people a year through 17,000+ healthcare workers.

20M

people reached with better care each year

17,000+

healthcare workers using the platform

countries across 5 world regions

Query response time, before → after

Local inference collapses a 5-minute wait into an instant answer.

5 min

Before

2 sec

After

Roughly a 150× faster response — near-instant at the point of care.

Token throughput (tokens / sec)

Answers now stream out about 22× faster.

0.83

Before

18.23

After

+26%

context recall vs. baseline

+32.5%

latency vs. MedLLaMA3-v20

Why it works in the field

Inference runs locally on edge devices, so the platform keeps working with no internet connection — answering natural-language questions in local languages.

Offline-capable Local-language answers Evidence-based responses Edge-deployed inference

Promise

Acts as a "second doctor" where physicians are scarce, moving beyond tag search to richer, natural-language insight — now piloting at six sites in The Gambia, Papua New Guinea and Ecuador.

Open challenges

Edge-hardware limits, ongoing verification of model accuracy, plus wider concerns of bias, privacy and compliance with local regulations.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.