NVIDIA dropped a new family of open AI models this week. The Nemotron 3 family — Nano, Super, Ultra — is built specifically for agentic, multi-agent workflows. The flagship Super model carries a 1 million token context window, which means it can hold an entire enterprise workflow in memory, understand the whole system, and execute across it without losing the thread.
That's not the story.
The story is who showed up first: Accenture. Deloitte. EY. CrowdStrike. ServiceNow. Palantir.
These are firms that built $10–100 billion businesses on a single premise: that human expertise, structured into projects and billed by the hour, is worth paying a premium for. They are, collectively, the most sophisticated human-intelligence delivery systems ever constructed.
And they just signed on as launch partners for open agentic AI.
171% Is the Number That Changed the Math
Enterprise data from April 2026 shows agentic AI delivering an average 171% ROI — three times traditional automation. Automation Anywhere reports AI agents auto-resolving 80%+ of IT support tickets and cutting ITSM costs by 50%. Microsoft's Dynamics 365 is shipping agentic ERP automation as a standard feature.
When Deloitte or Accenture sees those numbers, they don't run a pilot. They run an acquisition model.
The traditional consulting engagement: send in a team of 4–8 consultants, spend 90 days assessing and implementing, collect fees in the $500K–$2M range. The agentic alternative: deploy an agent that holds the entire system context in working memory, executes the assessment continuously, and flags issues in real time — at a fraction of the cost.
At 171% ROI, the math isn't subtle. It becomes a business model, not an experiment.
Why It Matters That Consultants Are First
Consulting firms are structurally conservative. They don't move until the risk of not moving exceeds the risk of moving. The fact that Accenture, Deloitte, and EY are named launch partners — not pilot users — means the ROI math has already cleared their internal hurdles.
These firms have armies of analysts. Junior associates. Tier-1 research roles. Data compilation work. Due diligence execution. Report synthesis. All of it sits directly in the capability range of a model with a 1-million-token context window and multi-agent orchestration built in.
They know this. That's why they're at the table early — because being early in a 171% ROI business is worth the reputational exposure.
What It Means for Your Hiring Strategy
If the companies that sell human expertise at premium rates are betting on agentic AI as their delivery model, the question for every CTO is: what work in your organization sits in the same category?
Junior-to-mid white-collar roles involving research, synthesis, report generation, tier-1 support, documentation, and data analysis — these are exactly what the consulting giants are automating first. Not because they're struggling to hire, but because the ROI math is unavoidable.
The Atlassian Pattern we covered this week — cut generalists, add AI specialists, at 2:1 — is exactly what Accenture and Deloitte are executing on now. Except they have the revenue scale to run it faster and the talent networks to hire for what comes next before the rest of the market figures out what's being built.
Companies that see this clearly are restructuring hiring before they're forced to. The ones that don't will run the same swap reactively — under board pressure, at higher cost, with worse candidates.
One Question
If Deloitte is deploying AI agents that do what their junior consultants do — at 171% ROI — what does that mean for the roles you're currently posting?
Because those models aren't staying inside consulting firms.
VC5 Consulting works with companies staffing for AI-native operations. If you're trying to figure out which roles on your org chart are on the displacement curve and what to hire instead — let's talk.