Jensen Huang walked on stage at GTC 2026 today and announced NemoClaw — NVIDIA's open-source enterprise AI agent platform built on OpenClaw. Pitched to Salesforce, Cisco, Google, Adobe, and CrowdStrike. Hardware-agnostic. Apache 2.0 licensed. Free to use.
The enterprise AI agent category just got its biggest validation moment in history.
And we couldn't be more excited about it.
Here's why — and here's the one thing NemoClaw doesn't solve.
---
What NemoClaw Actually Is
NemoClaw is an agent framework. A powerful one. It takes OpenClaw's foundation — the most-starred project on GitHub earlier this year — and adds enterprise-grade orchestration, authentication, and infrastructure optimizations on top.
Think of it like this: OpenClaw was the engine. NemoClaw is the full vehicle. Multi-agent workflows, built-in security features, pre-built connectors for major enterprise systems.
For enterprises that have been sitting on the sidelines while watching OpenClaw get employees banned from work devices — NemoClaw is the on-ramp. NVIDIA's credibility, enterprise focus, and hardware ecosystem make this a platform enterprises can actually evaluate seriously.
That's genuinely good for the entire industry.
---
The Problem That Didn't Get Solved Today
Here's what we kept thinking during the keynote.
A Meta employee recently shared a story about an OpenClaw agent going rogue and deleting her emails in bulk. A security researcher hijacked an OpenClaw agent in under two hours. Enterprises banned employees from using OpenClaw on work devices entirely due to unpredictable behavior.
NemoClaw addresses this by adding security features to the framework layer. Authentication. Access controls. Privacy tools.
That's necessary. But it's not sufficient.
Framework-level security tells your agents what they're allowed to access.
Execution-layer governance tells you what your agents actually did — and stops them before they do something wrong.
Those are fundamentally different problems.
NemoClaw is solving the first one. Nobody has solved the second one yet.
That's the gap.
---
The Four Questions NemoClaw Can't Answer
Every enterprise adopting NemoClaw will eventually face these questions:
1. What happens when an agent does something wrong?
With NemoClaw alone, you find out after the consequences. The email already sent. The record already modified. The data already accessed.
2. Who approved that action before it ran?
NemoClaw has no human-in-the-loop governance at the execution layer. There's no mechanism to pause an agent action, route it for human review, and resume only after approval.
3. Where's the audit trail when compliance asks?
Authentication logs tell you who accessed what. They don't tell you why an agent made a specific decision, which policy was evaluated, or what the agent was trying to accomplish when it took a risky action.
4. How do you stop it before the damage is done?
Framework security operates at the access control layer — it can prevent unauthorized access. It cannot intercept an authorized agent taking an unauthorized action in real time.
---
This Is the ServiceNow Moment for AI
ServiceNow didn't build SAP. They built the governance layer that made SAP deployable across the enterprise.
Every enterprise system needed workflow governance, audit trails, and human approval mechanisms. ServiceNow built that layer once and made it work above every system.
Today ServiceNow is worth $200B+.
The same dynamic is happening right now in enterprise AI.
NemoClaw, LangGraph, CrewAI, AutoGen — these are the SAPs of the agent era. Powerful frameworks for building autonomous AI.
Qoris is building the ServiceNow layer above all of them.
Not competing with NemoClaw. Governing it.
---
How Knox Closes the Gap
Knox is Qoris's execution-layer governance for enterprise AI agents.
Here's what it does that NemoClaw doesn't:
Intercepts Before Execution
Knox sits between your agents and every system they interact with. Before an agent sends an email, accesses a database, triggers a payment, or modifies a record — Knox evaluates the action against your defined policies.
Not after. Before.
Human-in-the-Loop Built In
When Knox flags a risky action, it pauses execution and routes to a human reviewer with full context — what the agent was trying to do, why it was flagged, what the downstream impact would be. Approve it and execution resumes. Deny it and the action is permanently blocked and logged.
Immutable Audit Trail
Every Knox evaluation is logged — timestamped, attributed, and immutable. Not just access logs. Decision logs. Every agent action, every policy evaluated, every human decision. The audit trail that makes AI deployable in regulated industries.
Framework-Agnostic
Knox connects via MCP — the Model Context Protocol. Any agent framework that supports MCP connects to Knox. NemoClaw included.
---
NemoClaw vs Qoris AI Workers
One more angle worth addressing directly.
NemoClaw is the right choice for enterprises with dedicated DevOps teams who want full control of their agent infrastructure and are comfortable with open-source maintenance.
Qoris AI Workers are the right choice for enterprises who want governed autonomous agents running this week without managing infrastructure.
| NemoClaw | Qoris AI Workers | |
|---|---|---|
| Deployment | Self-hosted | Cloud-native |
| Infrastructure | You manage | We manage |
| Time to production | 3-6 months | Days |
| Governance layer | DIY | Knox built in |
| Audit trail | Manual setup | Automatic |
| Compliance ready | Requires work | From day one |
These aren't competing products. They serve different buyers.
Many enterprises will end up using both — NemoClaw for engineering teams building custom agents, Qoris AI Workers for business teams deploying governed agents without the infrastructure burden.
---
The Bigger Picture
Today's GTC announcement didn't create competition for Qoris.
It created our market.
Every enterprise that adopts NemoClaw is an enterprise that needs Knox. Every organization deploying autonomous AI agents needs an execution-layer governance layer that can intercept bad actions before they happen, route decisions to humans when judgment is required, and produce an audit trail that compliance teams can actually use.
NVIDIA just made that market 10x larger.
The category is validated. The governance layer is ready.
Connect NemoClaw to Qoris via MCP →
---
*Eliel Joseph is the CEO and co-founder of Qoris AI — the trust layer for enterprise AI. Qoris builds Knox, the execution-layer governance for autonomous AI agents, and the Memory Engine, persistent versioned memory for any agent framework. Headquartered in Stamford, CT.*