Welcome back to EmergingTech.

Getting AI into hospitals has always stalled on one thing: patient data nobody will let you touch. Monash's new MAVERIC supercomputer is built specifically for that problem, giving health researchers a sealed environment to train models on real records without the privacy exposure.

Let’s dive in.

LATEST IN AI

OpenAI has launched LifeSciBench, a new benchmark designed to test how well AI models handle real world scientific research, moving beyond simple biology questions to complex research tasks and decision making.

Built with input from 173 scientists, the benchmark includes 750 expert created tasks across biotechnology, drug discovery, genomics, and other life science fields. Responses are evaluated using more than 19,000 grading criteria designed to mirror professional research standards.

The results show AI still has a long way to go. Even the top performing model passed just 36% of tasks, with performance dropping sharply when research involved complex data such as chemical structures, figures, and scientific documents.

Databricks has added Grok to its Agent Bricks platform, bringing the model into a governed environment where enterprises can build AI agents using their own data.

The integration, announced at the Databricks Data + AI Summit, also includes collaboration with Amazon Web Services via Bedrock, giving companies more flexibility to run frontier models alongside open source options where their data already resides.

The move reflects a push toward unified AI platforms where teams can choose models while keeping data, control, and governance in one place.

Sarvam AI has raised $234 million in the first close of its Series B round, valuing the company at $1.5 billion. New investors include HCL Technologies and Bessemer Venture Partners.

The startup says the funding will help accelerate development of its sovereign AI stack, spanning models, compute infrastructure, and deployments, as demand grows for locally controlled AI systems.

OpenAI says more than 230 million people now use ChatGPT each week for health and wellness questions, highlighting the platform’s growing role in healthcare. Led by researcher Karan Singhal, the company is working to make ChatGPT more accurate and reliable for medical guidance, with ambitions to play a bigger role in patient care. Read more→

AI health kit detects hidden TB - A 60 year old farmer in Karnataka discovered he had tuberculosis after an AI powered screening kit flagged the disease at a village health camp. The story, reported by The Better India, highlights how AI tools are helping bring early disease detection to rural India. Read more→

Monash launches secure AI supercomputer - Monash University has unveiled MAVERIC, a new AI supercomputer designed for health research using sensitive patient data. Powered by NVIDIA technology, the system gives approved researchers a secure environment to train AI models for projects ranging from skin cancer detection to epilepsy prediction. Read more→

Z.ai has expanded support for its new GLM 5.2 model across coding tools like Claude Code, OpenClaw, and Cline. The model supports up to 1 million tokens of context and is designed for more advanced coding and reasoning tasks.

How to use:

The update gives developers another powerful option as competition among AI coding assistants continues to heat up.