NVIDIA used the BoxWorks 2026 conference to show how its Nemotron open models, combined with business context and tools, can help developers build AI agents that make enterprise knowledge usable.
BoxWorks 2026 · NVIDIA
Nemotron open models aim to turn enterprise knowledge into working AI agents
At BoxWorks 2026, NVIDIA CEO Jensen Huang showed how Nemotron open models, paired with business context and tools, can help developers build agents that search, reason over and act on information an organization already holds.
How an enterprise agent is assembled
The announcement frames agent value as coming from connections, not from the model alone.
Step 1
Nemotron open models
Foundation for general AI capability
→
Step 2
Business context
Information the organization holds
→
Step 3
Tools
Means to carry out tasks
→
Result
Enterprise agent
Searches, reasons and acts on internal knowledge
The case for open
Developers can inspect, adapt and deploy the model within their own environments.
Suits enterprises handling sensitive internal data.
Models pair with business context and tooling companies already use.
Developers can start from these models to assemble knowledge-driven agents.
Open questions
Data quality and access control inside enterprise systems.
Reliability of multi-step tasks.
Cost of integration.
The announcement does not describe how these hurdles would be addressed in practice.
Key takeaway
The model is the foundation; the agent's value depends on the knowledge and tools it is connected to.
Not confirmed in the announcement
Model sizes
Benchmark results
Pricing
Availability dates
Named customer deployments
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...