Cassava AI integrates NVIDIA Nemotron to power local agentic AI

Expanding the AI Factory

Cassava Technologies has announced a significant expansion of its AI Factory capabilities. The company is integrating NVIDIA Nemotron, a series of open models, into its Cassava AI Multi-Model Exchange, known as CAIMEx. This platform provides a managed inference service, which is a way for companies to run AI models without having to build and maintain their own complex hardware and software infrastructure.

By offering these models as APIs, Cassava aims to give enterprises more flexibility. Instead of being locked into a single provider, businesses can choose the specific mix of models that best fit their unique needs. This approach is designed to help service providers and large companies grow their AI capabilities more affordably while maintaining control over their operations.

Powering Agentic AI

A primary goal of this deployment is to accelerate the adoption of agentic AI. Unlike standard chatbots that simply answer questions, agentic AI systems are designed to work through complex, multi-step tasks and interact directly with various business tools. NVIDIA Nemotron is specifically chosen for this because it brings efficient reasoning and tool-use capabilities to the CAIMEx platform.

Because these models are 'open,' they provide developers with access to model weights, training data, and specific recipes. This transparency allows engineers to fine-tune the models to create specialized agents. These agents can be tailored to specific company workflows, such as managing internal databases or automating complex administrative processes that require multiple actions over a long period.

Security and Governance

As AI agents begin to handle longer-running tasks and access internal business systems, security becomes a primary concern for many organizations. Cassava is addressing this by using NVIDIA OpenShell. This technology brings together model compute, governed agent execution, and independent security enforcement into a single environment. It ensures that while the AI is performing tasks, it remains within the boundaries set by the company.

This infrastructure is built to foster trust and confidence in data security. By hosting the models locally rather than in a public cloud, companies can ensure that their proprietary data remains within their own controlled environment. This is a critical requirement for industries that handle sensitive information, such as finance or healthcare, where data privacy is non-negotiable.

Optimizing AI Economics

One of the biggest hurdles for scaling AI is the cost of processing information. To solve this, Cassava introduced the CAIMEx Token Optimiser. This tool is designed to improve the economics of both generative and agentic AI by identifying information that has already been processed. When the system recognizes familiar data, it reuses it instead of processing the same content again.

This optimization reduces the total computing requirements for a given task. It helps customers use their AI tokens—the units of measurement for AI output—more efficiently. This benefit applies to both the models hosted by Cassava and any third-party models used on the platform. As a business scales its AI usage, this prevents unnecessary processing costs from spiraling out of control, making it easier to deploy AI at a larger scale.

The Path Forward

The integration of NVIDIA Nemotron is the latest step in Cassava's broader strategy to build an integrated AI ecosystem. By providing a combination of locally hosted infrastructure, diverse models, and specialized services, the company is creating a comprehensive toolkit for the modern enterprise. The goal is to provide the capacity and economics needed for businesses to move from simple AI experiments to full-scale production.

As businesses mature, the focus will shift toward how these agents can be integrated into everyday workflows. Cassava's move suggests that the next phase of AI adoption will be defined by how well these systems can be governed, secured, and made cost-effective enough for constant, high-volume use. By providing a managed way to access these capabilities, Cassava is positioning itself as a bridge between complex AI research and practical business application.