Nvidia is releasing three recent NIM microservices, or small independent services which can be a part of larger applications, to assist firms provide additional controls and security measures for his or her AI agents.
One of those recent NIM services is geared toward content security and stopping an AI agent from generating malicious or biased output. Another is designed to focus conversations only on approved topics, while the third recent service helps prevent jailbreak attempts by an AI agent or remove software restrictions.
These three recent NIM microservices are a part of Nvidia NeMo Guardrails, Nvidia's existing open source collection of software tools and microservices designed to assist firms improve their AI applications.
“By applying multiple lightweight, specialized models as guardrails, developers can close gaps that may arise when only more general global policies and protections are in place – because a one-size-fits-all approach doesn’t properly secure and govern complex agent AI workflows,” The Press release said.
It seems that AI firms are starting to comprehend that getting firms to adopt their AI agent technology won't be as easy as they first thought. While people like Salesforce CEO Marc Benioff recently predicted that greater than a billion agents will leave Salesforce alone in the following 12 months, the fact is prone to be a little bit different.
A recent study by Deloitte predicted that around 25% of firms are either already using AI agents or expect to achieve this in 2025. The report also predicted that by 2027, about half of firms will use agents. This shows that while firms are clearly excited about AI agents, they will not be adopting AI technology at the identical pace as innovation within the AI space.
Nvidia likely hopes that initiatives like these will make the introduction of AI agents seem safer and fewer experimental. Time will tell whether this is definitely true.
